V 0 5 M&R ARIES _ VOLUME 40 • PART 1 • MARCH 1997 Published by The Palaeontological Association • London Price £45-00 THE PALAEONTOLOGICAL ASSOCIATION (Registered Charity No. 276369) The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1996-1997 President : Professor D. Edwards F.R.S., Department of Earth Sciences, University of Wales College of Cardiff, Cardiff CF1 3YE Vice-Presidents : Dr J. A. Crame, British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 OET Dr P. D. Lane, Department of Earth Sciences, University of Keele, Keele, Staffordshire ST5 5BG Treasurer : Dr T. J. Palmer, Institute of Earth Studies, University of Wales, Aberystwyth, Dyfed SY23 3DB Membership Treasurer: Dr M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL Institutional Membership Treasurer: Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ Secretary: Dr M. P. Smith, School of Earth Sciences, University of Birmingham, Birmingham B15 2TT Newsletter Reporter: Dr S. Rigby, Department of Geology and Geophysics, Grant Institute, West Mains Road, Edinburgh EH9 3JW (co-opted) Marketing Manager : Dr A. King, English Nature, Northminster House, Peterborough PEI 1UA Publicity Officer: Dr M. A. Purnell, Department of Geology, University of Leicester, University Road, Leicester LEI 7RH Newsletter Editor : Dr R. B. Rickards, Department of Earth Sciences, University of Cambridge, Cambridge CB2 3EQ Editors Dr D. M. Unwin, Geology Department, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ Dr P. Doyle, School of Earth Sciences, University of Greenwich, Brewhouse Lane, London El 9NU Dr D. A. T. Harper, Department of Geology, University College, Galway, Ireland Dr C. J. Cleal, Department of Botany, National Museum of Wales, Cardiff CF1 3NP Dr R. M. Owens, Department of Geology, National Museum of Wales, Cardiff CF1 3NP Dr B. M. Cox, British Geological Survey, Keyworth, Nottingham NG12 5GG Dr D. K. Loydell (Technical Editor), Department of Geology, University of Portsmouth, Burnaby Building, Burnaby Road, Portsmouth POl 3QL Other Member : Dr R. Wood, Department of Earth Sciences, University of Cambridge, Cambridge CB2 3EQ Overseas Representatives Argentina : Dr M. O. Mancenido, Division Paleozoologia invertebrados, Facultad de Ciencias Naturales y Museo, Paseo del Bosque, 1900 La Plata. Australia: Dr K. J. McNamara, Western Australian Museum, Francis Street, Perth, Western Australia 6000. Canada: ProfessorS. H. Williams, Department of Earth Sciences, Memorial University, St John’s, Newfoundland A1B 3X5. China: Dr Chang Mee-mann, Institute of Vertebrate Palaeontology and Palaeoanthropology, Academia Sinica, P.O. Box 643, Beijing. Dr Rong Jia-yu, Nanjing Institute of Geology and Palaeontology, Chi-Ming-Ssu, Nanjing. France : Dr J.-L. Henry, Institut de Geologie, Universite de Rennes, Campus de Beaulieu, Avenue du General Leclerc, 35042 Rennes Cedex. Iberia: Professor F. Alvarez, Departamento de Geologia, Universidad de Oviedo, C/. Jesus Arias de Velasco, s/n. 33005 Oviedo, Spain. Japan : Dr I. Hayami, University Museum, University of Tokyo, Hongo 7-3-1, Tokyo. New Zealand: Dr R. A. Cooper, New Zealand Geological Survey, P.O. Box 30368, Lower Hutt. Scandinavia : Dr R. Bromley, Fredskovvej 4, 2840 Holte, Denmark. USA : Professor A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66044. Professor N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403. Professor M. A. Wilson, Department of Geology, College of Wooster, Wooster, Ohio 44961. Germany: Professor F. T. Fursich, Institut fur Palaontologie, Universitat, D8700 Wurzburg, Pleicherwall 1 MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1 997 are : Institutional membership . . £90-00 (U.S. $175) Student membership .... £1 1-50 (U.S. $20) Ordinary membership . . £28-00 (U.S. $50) Retired membership .... £14-00 (U.S. $25) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ. Student members are persons receiving full-time instruction at educational institutions recognized by the Council. On first applying for membership, an application form should be obtained from the Membership Treasurer: Dr M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL. Subscriptions cover one calendar year and are due each January ; they should be sent to the Membership Treasurer. All members who join for 1997 will receive Palaeontology, Volume 40, Parts 1-4. Enquiries concerning back numbers should be directed to the Marketing Manager. Non-members may subscribe, and also obtain back issues up to five years old, at cover price through Blackwell Publishers Journals, P.O. Box 805, 108 Cowley Road, Oxford OX4 1FH, UK. For older issues contact the Marketing Manager. US Mailing: Second class postage paid at Rahway, New Jersey. Postmaster: send address corrections to Palaeontology , c/o Mercury Airfreight International Ltd, 2323 EF Randolph Avenue, Avenel, NJ 07001, USA (US mailing agent). Cover: SEM of sporangium of Cooksonia pertoni ssp. apiculispora from Brown Clee Hill, Shropshire. This is the specimen (NMW 94 60G) that confirmed the vascular status of Cooksonia ; x 70. Photograph published originally in Nature, 357, 683-685, figure la. THE PERMIAN CORAL NUM IDIAPHYLLUM: NEW INSIGHTS INTO ANTHOZOAN PHYTOGENY AND TRIASS1C SCLERACTINIAN ORIGINS by Y. EZAKI Abstract. The Permian coral Numidiaphyllum, having an unusual septal arrangement and an aragonitic skeleton, has been classified in the Rugosa. The type species of Numidiaphyllum shows high intraspecific morphological variability and distinct granulation on septal faces. Ontogenetic development indicates that corallites show hexameral septal arrangement and cyclic mode of insertion. The genus has no specific morphologies that deny scleractinian affinities. Numidiaphyllum is believed to have originated in sponge-algal reefs in the Permian tropics but possesses a basic scleractinian form which was already established in Early Palaeozoic times. Whatever their variation might be, the zoantharians, which may be closely related to Numidiaphyllum , survived the end-Permian extinction in ‘deep-water' refuges as Permian holdovers, retaining their body plan, and they are possible scleractinian ancestors in the Triassic. Scleractinia have no immediate phylogenetic relationship to Rugosa. This study provides evidence about Permo-Triassic anthozoan phytogeny in terms of Permian survivors and their relationship to Triassic scleractinian origins. Cnidaria are among the oldest phyla of eumatozoans, and some constituent groups with calcareous skeletons (corals) have contributed much to reef construction throughout geological time. The Rugosa originated in the mid Ordovician and became extinct by the end of the Permian (Hill 1981), whereas the Scleractinia have been thought to have appeared in the Triassic and now play an important role in reef construction. Corals of the Rugosa have a calcitic skeletal mineralogy and serial insertion of septa at four specific sites during their ontogeny. The Scleractinia have an aragonitic mineralogy and cyclic insertion of septa within sextants (Oliver 1980). Extinction of the Rugosa and the origin of the Scleractinia have attracted much attention, and are often discussed together. Various approaches to understanding the latter have been made, especially with Permian rugosans and Triassic scleractinians, using the criteria of septal arrangement, mode of septal insertion, fine skeletal structure, mineral composition and biostratigraphical evidence (e.g. see Schindewolf 1942; Iljina 1965, 1984; Cuif 1980; Oliver 1980; Ezaki 1989). For recent reviews, see Oliver (1980), Iljina (1984) and Sando (1993). Many works have expanded on two conflicting views: (1) Rugosa were direct ancestors of Scleractinia (‘direct-origin hypothesis’) represented by Schindewolf (1942), Iljina (1965, 1984) and Cuif (1980); (2) Scleractinia originated independently of Rugosa (‘independent-origin hypothesis') supported by Oliver (1980), Hill (1981) and Ezaki (1989). However, the issue remains unresolved because most interpretations have depended on circumstantial evidence such as apparent similarities or differences in morphologies and structural patterns. The absence of corals in the Lower Triassic as well as apparent differences in the ranges of Rugosa and Scleractinia have supported the idea of the independent origin of Scleractinia (Oliver 1980). Whatever the scleractinian origin might be, both hypotheses have the common view that the Scleractinia appeared first in the Mid Triassic. Since their appearance, they have diversified into a range of environments, occasionally forming large-scale organic reefs. Four microstructural groups (fascicular, thick-trabecular, minitrabecular and pachythecal) are recognized in the Early Mesozoic Scleractinia, three of which are already found in the lower Middle Triassic (Anisian), suggesting ‘polyphyletic’ origins (Roniewicz and Morycowa 1993). However, molecular phylogenetic study of [Palaeontology, Vol. 40, Part 1, 1997, pp. 1-14, 1 pl.| © The Palaeontological Association PALAEONTOLOGY, VOLUME 40 2 the Anthozoa indicates that at least extant Scleractinia are monophyletic (Chen et al. 1995). Recently, a new scleractinian-like coral ( Kilbuchophyllia ) was found in the Ordovician rocks of the Southern Uplands, Scotland, and a new order Kilbuchophyllida was proposed in the Zoantharia to accommodate it (Scrutton and Clarkson 1991; Scrutton 1993). This material was interpreted as indicating a long Palaeozoic history for the anemone group, the Corallimorpharia, from which the Scleractinia were considered to have evolved in the Triassic. The phylogenetic significance of this in terms of anthozoan evolutionary patterns should attract considerable attention. The genus Numidiaphyllum comprises Permian corals, having a scleractinian-like septal development (Fliigel 1976) and an aragonitic skeleton (Wendt 1990a). Both these authors placed the genus in the Rugosa, with rugosan septal notation providing evidence for a direct-origin hypothesis. With proper understanding of its morphological characteristics and variability, it is now possible to determine whether the genus really belongs to the Rugosa. This paper presents the ontogenetic development and intraspecific morphological variability of the type species of Numidiaphyllum with special analysis of the mode of septal insertion. The evaluation of the higher taxonomic position of the genus gives new insights into anthozoan phylogeny, especially that of the Zoantharia. This study also explains the mutual relationship between the end-Permian extinction of corals and a possible Triassic scleractinian origin. MATERIALS AND FOSSIL OCCURRENCE Mid Permian limestones, containing a highly diverse fauna and flora of calcisponges, algae, bryozoans and fusulinids ( Neoschwagerina-Yabeina assemblage), are found in the Djebel Tebaga area of southern Tunisia (Newell et al. 1976). The Permian units are characterized by a reef complex, consisting of large bioherms, and are overlain by thick non-marine red-beds. Colonial coralla of Numidiaphyllum have been collected from Section C, west of Merbah el Oussif; the material comes from Bed 19, about 7 m thick, of the Upper Biohermal Complex ( Yabeina Zone) of the Djebel Tebaga Reef (Newell et al. 1976), and was first described by Fliigel (1976). Corals other than Numidiaphyllum have been described by Stevens (1975) and Termier et al. (1977). The corallum is entirely embedded in the limestone matrix. In some cases, erosion before burial has destroyed parts of the outer wall. Corallites are rimmed with a dark micrite layer and encrusted with bryozoans and algae, forming a coral bafflestone. The coral skeleton is partly dissolved and filled with coarse, sparry calcite so that inner structures are obliterated and broken. Elsewhere, an interstitial matrix consists of brick- to yellow-coloured carbonate with irregular fenestral fabrics. More than 40 corallites were available for study. Skeletal debris includes sponges, algae, bryozoans, fusulinids, crinoid fragments, gastropods, bivalves and smaller foraminifera. SYSTEMATIC PALAEONTOLOGY The standard descriptive terminology for the Scleractinia (e.g. Wells 1956) is used in this account but, for better understanding of morphologies, the rugosan equivalents (e.g. Hill 1981) are indicated when necessary. A protocorallite cannot be detected in the material available, so knowledge of ontogeny (hystero-ontogeny) and septal development is based entirely on asexually reproduced corallites (hystero-corallites). Although offsets, which inherit septa from the parent, do not show the EXPLANATION OF PLATE 1 Polished surface of Numidiaphyllum gillianum Fliigel, 1976; corallum is fasciculate and phaceloid. Corallites are widely spaced; Middle Permian ( Yabeina Zone), west of Merbah el Oussif, Djebel Tebaga, southern Tunisia; x 1. 1, USNM 248224, 248229-30, 248235-36; 2, USNM 248223, 248227, 248231-34, 248237-38. For key to numbering, see Text-figure 1. PLATE 1 EZAKI, Numidiaphyllum 4 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Correspondence of specimens figures in Plate 1 with registration numbers and corallites shown in Text-figures 2-3. a, 1 : USNM 248229 (Text-fig. 2f), 2: USNM 248236 (Text-fig. 2b), 3: USNM 248235 (Text-fig. 2c), 4: USNM 248230 (Text-fig. 2d), 5: USNM 248224; b, 1: USNM 248237 (Text- fig. 3d), 2: USNM 248227, 3: USNM 248238 (Text-fig. 3e), 4: USNM 248223 (Text-fig. 3b), 5: USNM 248233 (Text-fig. 3c), 6: USNM 248231 (Text-fig. 3a), 7: USNM 248234, 8; USNM 248232. initial stage of septal insertion, subsequent development is subject to the septal pattern inherent in the coral (Hill 1981). A dorso-ventral direction is not clear in the type species of Numidiaphyllum because ( 1 ) no protocorallites are found, and initial modes of septal insertion are unknown ; (2) septa apparently show variable arrangement, although it is essentially hexameral; (3) there is no elongate or blade-like columella to indicate a plane of bilateral symmetry. Repository of specimens. The specimens are now stored in the US National Museum of Natural History, Washington, D.C., USA (USNM). Phylum cnidaria Hatschek, 1888 Class anthozoa Ehrenberg, 1834 Subclass zoantharia de Blainville, 1830 Order scleractinia Bourne, 1900 Family numidiaphyllidae Fliigel, 1976 Genus numidiaphyllum Fliigel, 1976 Type species. Numidiaphyllum gillianum Fliigel, 1976. Diagnosis. Corallum solitary(?) and fasciculate, phaceloid with aragonitic skeleton; corallites cylindrical and reproduced by intratentacular increase; walls broadly indented; six first cycle septa EZAKI: PERMIAN CORAL 5 text-fig. 2. Numidiaphyllum gillianum Fliigel, 1976; transverse features of corallites; Middle Permian ( Yabeina Zone), west of Merbah el Oussif, Djebel Tebaga, southern Tunisia; x T9. A, USNM 248226; B, USNM 248236; c, USNM 248235; d, USNM 248230; E, USNM 248239; F, USNM 248229; black dots represent the six first cycle septa. show hexameral arrangement; septa arranged radially and radiobilaterally ; first order septa vary in length, and higher order septa may appear sporadically; septa granulate and showing orthogonal fine structure; neither dissepiments nor columella present. Range. Middle Permian ( Yabeina Zone in fusulinid zonation). Geographical distribution. Djebel Tebaga, southern Tunisia. Remarks. In his original description, Fliigel (1976) referred to the morphological similarity between Numidiaphyllum and polycoeliid corals within the Rugosa, and placed Numidiaphyllum in his new family Numidiaphyllidae within that order. Hill (1981) doubted the suprageneric classification proposed by Fliigel (1976) and preferred to assign Numidiaphyllum to Subclass uncertain, whereas 6 PALAEONTOLOGY, VOLUME 40 text-fig. 3. Numidiaphyllum gillianum Fliigel, 1976; transverse features of corallites; note a marked variation in septal arrangement; for further explanation, see text; Middle Permian (Yabeina Zone), west of Merbah el Oussif, Djebel Tebaga, southern Tunisia; x T9, a, USNM 248231; b, USNM 248223; c, USNM 248233; d, USNM 248237; e, USNM 248238. Yii et al. (1983) and Iljina (1984) placed it respectively in the rugosan subfamily Plerophyllinae and in the Nunudiaphylhdae. Wendt (1990n) considered that Numidiaphyllum usually showed ‘serial’ septal insertion typical of the Rugosa, and placed the genus in the rugose family Polycoeliidae. He also suggested the primary, aragonitic composition of this ‘rugosan’ genus (see later) which has since been cited as strong evidence for a possible phylogenetic relationship between Palaeozoic Rugosa characterized by a calcite skeleton and the post-Palaeozoic aragonitic Scleractinia, supporting the idea of direct scleractinian origin. Nudds and Sepkoski (1993) placed Fliigel’s (1976) Numidiaphyllidae in Order uncertain. The present work will demonstrate that the genus Numidiaphyllum has no specific morphologies that deny scleractinian affinities, and that it is a Palaeozoic representative of the order Scleractinia, not the Rugosa. Numidiaphyllum gillianum Fliigel, 1976 Plate 1 ; Text-figures 2-7 Holotype. USNM 248220, from the Middle Permian ( Yabeina Zone), west of Merbah el Oussif, Djebel Tebaga, southern Tunisia; para types. USNM 248221-41, horizon and locality as above. Diagnosis. Corallum solitary!?) and fasciculate, phaceloid by intratentacular tabularial increase; septa in four orders and arranged radially and radiobilaterally ; septa variable in length in later EZAKI: PERMIAN CORAL 7 text-fig. 4. a, daughter corallites produced by intra tentacular tabularial increase; x 1-2; b, close-up of a daughter corallite shown in a, indicating a hexameral septal arrangement and sporadically formed, fourth order septa, USNM 248221, x 3 3. text-fig. 5. Polished surface of Numidiaphyllum gillianum Fliigel, 1976, showing granulate septa on lateral surfaces; x 2. stages and higher order septa less well developed; septa variably granulate; no dissepiments or columella. Description. The corallum is solitary!?) and fasciculate, phaceloid (PI. 1). Corallites are cylindrical, ranging from 8-9 to 38 mm in diameter (average 20 mm), and widely spaced without connecting processes. About ten corallites occur within 0 01 square metres in transverse section. No corallite outer-surfaces are observed. The wall is generally thin and broadly embayed at joints with septa (Text-figs 3b-c, 6m-n). Asexual reproduction occurs by means of tabularial increase (Text-figs 2f, 4a). Corallites are circular to elliptical in transverse section. Septa are in four orders, but the fourth order septa appear only in some loculi. Septa are arranged radially, essentially showing a hexameral arrangement with six first cycle septa (Text-figs 2b, d, 4b, 7). A radio- bilateral symmetry is also achieved by an undeveloped protoseptum and/or accelerated or retarded insertion of septa (Text-fig. 3c-d). There are 18 septa in a corallite at a diameter of 19 mm. The first order septa taper axially leaving an open space, but may be spindle-shaped, tapering towards the wall. They are slightly llexuous and occasionally conjoined axially without forming a columella. No specific septum is lengthened preferentially. Apart from the six first cycle septa, the septa may be long and almost equal in length. The second order alternates with the first order, though not always, and is variable in length up to 80 per cent, of the corallite radius. The third order septa originate between neighbouring first and second order septa. The fourth order septa occur sporadically and are present as ridges (Text-figs 4b, 7c). Septa are granulate on lateral surfaces (Text-figs 2a, 5) and are rarely carinated. The dilated, first order septa delimit parricidal daughter corallites (Text-fig. 2f). The fine skeletal structure is obscure due to recrystallization, although relics of orthogonal (fibro-normal) structure may be traced near the septal periphery (Wendt 1990n, fig. 1-4; 1990fi, fig. 6e). Tabulae are complete and irregularly spaced. They are concave upwards to various degrees, but may be slightly convex peripherally (Text-fig. 2e). Five to seven tabulae are counted in a vertical distance of 10 mm. Transversely cut edges of tabulae are seen in interseptal spaces, showing a concentric and herringbone arrangement. Neither dissepiments nor columella are present. PALAEONTOLOGY, VOLUME 40 text-fig. 6. Numidiaphyllum gillianum Fliigel, 1976. Serial transverse features of the holotype, showing ontogenetic development of septa and septal arrangement. Note a high variability in length of second order septa and walls embayed at joints with septa. For further explanation, see text; Middle Permian (Yabeina Zone), west of Merbah el Oussif, Djebel Tebaga, southern Tunisia, USNM 248220; x 1-8. Mode of asexual reproduction and ontogeny. Offsets originate in the tabularial part, and some of the first order septa become the dividing walls of daughter corallites, together with the walls formed at the axis (Text-fig. 2f). Parricidal offsets are separated and replace the parent corallite. Each daughter corallite inherits the old septa of the parent corallite (atavo-tissue) at its periphery and adds new septa of its own (neo-tissue) on the opposite side. Text-figure 6 shows the ontogenetic EZAKI: PERMIAN CORAL 9 morphological changes of a daughter corallite which was reproduced by means of tabularial increase. In cases of tabularial increase, septal length in the daughter corallite is somewhat irregular at first. The septa are less developed on the dividing walls and in the atavo-tissue (Text-figs 2f, 6a). They are not differentiated in length and thickness in the neo-tissue (Text-figs 2f, 6a-b), where transversely cut edges of tabulae are present in a concentric arrangement. The second order septa are inserted between the first order septa, and the third order septa appear between the undifferentiated first order and second order septa (Text-fig. 6c). The septa taper axially with a thickened corallite wall. At a later stage (Text-fig. 6d), they are much differentiated in length, though variable, taking on an apparently radial symmetry. The third order septa then disappear and are restricted to the wall (Text-fig. 6e). In the following stage (Text-fig. 6f), seven septa are developed and, when present in an interseptal space, the second order septa are short. As a corallite grows, it maintains a similar appearance with its essential features unchanged but with second order septa of variable length. An almost completely radial arrangement appears with the axial elongation of the second order septa (Text-fig. 6o). The third order septa occur sporadically as low ridges. Septal insertion. By tracing each individual septum in serial transverse sections, allowing for the effects of septal stunting (see Weyer 1972), the mode of septal insertion (timing, location and order) can be determined. It is difficult to discern the initial mode of septal insertion in N. gillianum because daughter corallites are produced asexually by intratentacular tabularial increase. Fliigel (1976) mentioned that metasepta of N. gillianum originated in an irregular manner. Wendt (1990a) considered that septa were produced ‘serially’ in a polycoeliid manner characteristic of the Rugosa. However, the ‘first pair of metasepta’ (Wendt 1990a, fig. 2-4) are not the first appearing metasepta (sensu Rugosa), but part of the higher orders of septa ( sensu Scleractinia), although the septa designated as cardinal, counter, two alar and two counter-lateral correspond to the first cycle septa. In N. gillianum , the fourth order septa are present sporadically as low ridges (Text-figs 2a, 4b). Six first cycle septa are present, and the six septa divide the corallite into sextants for subsequent septal insertion, resulting in a hexameral septal arrangement (Text-fig. 7). Although the daughter corallites inherit some septa from the parent corallites and septa are occasionally retarded and/or accelerated in insertion, higher orders of septa, up to the fourth order, appear between neighbouring lower orders of septa in each sextant. Hence the second order septa are inserted between the first order septa, and third order septa appear between first order and second order septa in succession (Text-figs 6b-d, 7d-f). These features of septal arrangement and insertion strongly indicate an affinity close to the Scleractinia because the Scleractinia have six protosepta regulating the locations of subsequent septal insertion in cyclic fashion (Wells 1956). Variability. Growth forms of the corals are solitary(?) and fasciculate, phaceloid. In corallites with a fourth order of septa, corallite diameter varies, ranging from 8-9 to 38 mm. Septal insertion is apparently incomplete because some higher order septa do not appear and are restricted to the wall. Septa are highly variable in length throughout corallite growth. The fourth order septa occur only sporadically. The timing and degree of septal stunting and withdrawal, especially of higher order septa, may be related to the formation of tabulae. Septa are granulate on lateral surfaces in some corallites (Text-figs 2a, 5) but are smooth in other corallites. Septa are variably dilated, tapering and fusiform. Septal arrangement is essentially hexameral, with six first cycle septa (Text-figs 2b, 7f), and is more evident in younger corallites (Text-figs 2d, 4b). However, this varies among individuals, especially in later stages. Septa are not always developed equally within a corallite. Corallites take on tetrameral (Text-fig. 3e) and pentameral structure showing a bilateral symmetry (Text-fig. 3c-d), due to an underdevelopment of first order septa and/or a retardation of septal insertion on one side. Later, this pattern may be hexameral with the introduction of the higher orders of septa in retarded 10 PALAEONTOLOGY, VOLUME 40 text-fig. 7. Numidiaphyllum gillianum Fliigel, 1976. Serial transverse features of daughter corallite, showing ontogenetic development of septa and septal arrangement. Septa may show a bilateral symmetry but are later characterized by a hexameral symmetry. Septa are highly variable in length throughout growth. Black dots represent the six first cycle septa. For further explanation, see text; magnification unknown (modified after Wendt 1990a). sextants (Text-fig. 7f). Others seemingly show heptameral (Text-fig. 6) or octameral structure, with one or two additional, elongated septa probably of second cycle origin. The septal arrangement is thus variable throughout ontogenetic (growth) stages. Septal dilation at the time of intratentacular tabularial budding may also be related to the apparent variation in septal arrangement. A corallite that is elongate and/or elliptical in outline, resulting from oblique sectioning, original shape, or compaction commonly shows septal acceleration in the end sextants (Wells 1956, fig. 240b). When intratentacular increase occurs, offsets are separated by dilated first order septa of the parent corallite and by new walls in the tabularium. The parent corallites are divided into several daughter corallites which are outlined irregularly at first (Text-figs 2f, 4a), with septa not clearly differentiated in length and thickness. Septal insertion is retarded on the dividing walls and in the atavo-tissue. Some variability, including septal length, arrangement and thickness, is also induced by intratentacular asexual reproduction. Remarks. N. gillianum is similar to the Triassic scleractinian Stylophyllopsis zitteli described originally by Freeh (1890) from the Rhaetian of the Zlambach Beds at Fischerwise, Austria. Both species show hexameral septal arrangement, especially in younger corallites, and granulation of septa. Seemingly large morphological variability in septal length, thickness and development is also commonly observed. However, N. gillianum lacks septal spines, even in the calicular part, as well as dissepiments. It shows an orthogonal fine structure of septa (Wendt 1990a, 19906), but its relationship to the fascicular type characteristic of Stylophyllopsis (see Cuif 1972; Roniewicz and Morycowa 1993) is uncertain. N. gillianum is akin to the Triassic scleractinians Volzeia fritschi and Protoheterastraea leonhardi in hexameral septal arrangements and septal development during the course of ontogeny. However, it is separable from these two species by sporadically distributed higher orders of septa and the absence of dissepiments. EZAKI: PERMIAN CORAL SKELETAL COMPOSITION AND PA L AEOEN VI RON M EN TS The skeleton of the corals here studied is composed of a fine- to coarse-grained mosaic of neomorphic calcite interpreted as being converted from aragonite. The septa may be rimmed with a micritic envelope cemented by rinds of acicular calcite. Interseptal voids are filled with coarse, mosaic spar (Wendt 19906, fig. 5e-f). Wendt (1990a) showed high Sr++ values in calcitized Numidiaphyllum skeletons from Tunisia, and the present author follows Wendt’s (1990a, 1990 b) conclusion that the original skeleton of this genus was aragonitic, although X-ray diffraction analysis shows no aragonite peak. In Tunisia, aragonitic mineralogies are still preserved in some reef components (Wendt 1977). Aragonite precipitation from seawater is kinetically favoured at higher temperature, higher Mg++ and lower Pco2 levels, and other complex chemical conditions (Tucker and Wright 1990, p. 409). In the Mid Permian limestones of southern Tunisia, Numidiaphyllum occurs where a highly diverse biotic association dominated by calcisponges and algae has been recognized (Newell et al. 1976; Senowban-Daryan and Rigby 1988). A specific biotic community, shown by a calcisponge- algal association, was apparently crucial to the first appearance and survival of Numidiaphyllum which lived both as a dweller and baffler in warm-water sponge-algal reefs in tropical latitudes. SCLERACTINIAN AFFINITY OF NUMIDIAPHYLLUM On both age and mineralogical criteria, Numidiaphyllum could be considered ( 1 ) as merely a variant within the Rugosa, or (2) as a precocious scleractinian. Septa are marked on the corallite wall by distinct indents (Text-figs 3c, 4b, 6m-n), like the septal grooves which characterize Rugosa, not by costae as typify Scleractinia. However, such septal indents are also found in the scleractinian Stylophyllopsis mucronata from the Lower Jurassic of Britain (Duncan 1867). Septal faces are granulate to various degrees. Six first cycle septa divide the corallite in sextants, and subsequent second and third cycle septa are inserted in each interseptal space, showing a hexameral, cyclic insertion. The mode of septal insertion, the most important character for judging suprafamilial position, is like those in the Scleractinia. Intratentacular tabularial increase is a mode of asexual reproduction, and a similar mode is found not only in Rugosa (e.g. Entelophyllum articulatum ), but also in Scleractinia (e.g. Stylophyllopsis rudis and Volzeia fritschi). The weight of evidence indicates that Numidiaphyllum is neither a rugosan variant, regardless of its occurrence in the Permian, nor a representative of an intermediate group between Rugosa and Scleractinia; the latter has been erroneously suggested for Permian corals showing occasional irregular septal insertion (Schindewolf 1942; Iljina 1965, 1984). See Oliver (1980) and Ezaki (1989), for further discussion. The question to be addressed is therefore what is the higher taxonomic position of this Permian scleractinian-like coral with an aragonitic composition. ANTHOZOAN PHYLOGENY AND TRIASSIC SCLERACTINIAN ORIGINS Palaeozoic scleractinian-like corals have been reported from the middle Ordovician (Erina and Kim 1980; Scrutton and Clarkson 1991; Scrutton 1993), and the genus Kilbuchophyllia shows scleractinian affinity on the basis of latex replicas of the microarchitecture of the skeletal elements and septal pattern (Scrutton and Clarkson 1991). Kilbuchophyllia , in the new order Kilbucho- phyllida, was interpreted by those authors as being an early example of skeletonization from among a group of sea-anemones which were possible ancestors of the Triassic scleractinians. Apart from Kilbuchophyllia , Palaeozoic scleractinian-like corals so far recognized were reviewed briefly by Scrutton and Clarkson (1991) and shown to be unacceptable or too poorly known. Kilbuchophyllia is the only suitable Palaeozoic subject for this comparison. It must be decided whether 12 PALAEONTOLOGY, VOLUME 40 Numidiaphyllum, with a basic, scleractinian body plan, should be allied to the Kilbuchophyllida or the Scleractinia. Scleractinian-like corals occur in both the Ordovician and Permian, showing the persistence of zoantharians with a scleractinian body plan, even in the barren intervals during Silurian to Carboniferous times, a time duration of at least 150 My. The absence of undoubted scleractinian- like corals for such a long period and, moreover, a close morphological similarity to some Triassic scleractinians (see remarks in Systematic Palaeontology and Scleractinian Affinity of Numidia- phyllwn ), suggest a more likely placement of the Permian Numidiaphyllum with the Scleractinia rather than the Kilbuchophyllida. A large number of skeletonized benthic organisms became extinct at the end of the Permian in the severest extinction of Phanerozoic time (e.g. Sepkoski 1989; Erwin 1993, 1994), and no Early Triassic corals are known. It may seem surprising that the new genus Numidiaphyllum originated during the late Mid Permian, when rugose and tabulate corals greatly declined phylogenetically into the end-Permian extinction. The Late Permian decline and disappearance of the Rugosa occurred via morphological changes, from complex to simple forms, reflecting phylogenetic trends and differential deteriorating environments (Ezaki 1993, 1994). Fasciculate rugose corals survived up to the latest Permian Changxingian but became extinct by the end of the period (Ezaki 1994). The appearance of Numidiaphyllum with an aragonitic skeleton, might not have an immediate causal relationship with the decline of rugose and tabulate corals. It lived in exceptionally favourable warm-water conditions. No evidence has been found of reefs in the Early Triassic oceans, when protracted inhibiting conditions for reef-forming organisms prevailed (Stanley 1988). The Permian survivors, characterized by calcisponges and algae, were considered to have played a crucial role in the re- establishment of reefs in the Mid Triassic after a period of global reef-gap (Stanley 1988; Hallam 1991). If the Scleractinia did not possess zooxanthellae until the Late Triassic (Stanley and Swart 1995), azooxanthellate zoantharians with a basic, scleractinian body plan could have grown in ‘deep-water’ settings during the Early Triassic reef-gap period. Whatever their variation might be, the zoantharians, which may be closely related to Numidiaphyllum , survived the disastrous end- Permian extinction in refuges. With global removal of inhibiting factors everywhere in the Anisian ocean (see Ezaki 1995), as well as regional amelioration of substrate conditions, scleractinian corals dominated in association with calcisponges and algae. Geochemical seawater conditions also favoured aragonite-secreting organisms during Triassic times (Railsback and Anderson 1987). The difference in basic body plan between the Rugosa and the Scleractinia was not established around the time of the Permian-Triassic boundary but already in Early Palaeozoic times. Whatever the origin, Scleractinia have no immediate phylogenetic relationship to Rugosa, arguing against the direct-origin hypothesis. CONCLUSIONS The end-Permian extinction played a significant role in anthozoan evolutionary patterns because the Rugosa and the Tabulata became extinct. However, the zoantharians with a basic, scleractinian body plan survived and gave rise to Triassic scleractinian ancestors. The taxonomic reinterpretation of the Permian coral Numidiaphyllum provides a clearer scenario for anthozoan phylogeny in terms of the end-Permian extinction and the origin of Triassic Scleractinia. Acknowledgements . I express my gratitude to Dr C. T. Scrutton (University of Durham) for invaluable discussions and suggestions during this study. Dr B. R. Rosen (The Natural History Museum, London) made useful comments on scleractinian corals. I thank Professor Emeritus M. Kato (Hokkaido University), Drs W. A. Oliver, Jr and W. J. Sando (US Geological Survey), and Dr J. W. Pickett (Geological Survey of New South Wales) for reviewing an earlier manuscript. Dr W. A. Oliver, Jr and Ms. J. Thompson (US Geological Survey) permitted me to examine the specimens in their care. This study was supported by the Scientific EZAKI: PERMIAN CORAL 13 Research Fund from the Japanese Ministry of Education, Science and Culture (nos 03041069, 04304009, 05304001). REFERENCES blainville, h. m. d. de 1830. Zoophytes. Dictionaire de Sciences Naturelle de Paris , 60, 1 -546. bourne, G. c. 1990. Anthozoa. In landkester, r. (ed.). Treatise on Zoology , II. London. chen, c. a., odorico, d. m., lohuis m. ten, veron, J. E. N. and miller, d. j. 1995. Systematic relationships within the Anthozoa (Cnidaria: Anthozoa) using the 5'-end of the 285 rDNA. Molecular Phylogenetics and Evolution , 4, 175-183. cuif, J. p. 1972. Recherches sur les Madreporaires du Trias. I. Familial des Stylophyllidae. Bulletin du Museum National d Histoire Naturelle , 97, 211-291. — 1980. Microstructure versus morphology in the skeleton of Triassic scleractinian corals. Acta Palaeontologica Polonica , 25, 361-374. duncan, p. M. 1867-68. Supplement to the fossil corals. Part 4, Liassic. Monograph of the Palaeontographical Society , Part 1, 20 (85), l-iii, 1-44, pis 1-11 ; Part 2, 21 (90), 45-73, pis 12-17. ehrenberg, c. G. 1834. Beitrage zur physiologischen Kenntnis der Corallenthiere im allgemeinen, und besonders des rothen Meeres, nebst einem Versuche zur physiologischen Systematik derselben. Koniglischen Akademie der Wissenschaft , Physiologische-Mathematischen Abhandlung , 1832, 225-380. erina, M. v. and kim, a. I. 1980. On some Ordovician Scleractinia-like corals from the south Tien-Shan. Acta Palaeontologica Polonica , 25, 375-379. erwin, D. h. 1993. The great Paleozoic crisis. Columbia University Press, New York, 327 pp. 1994. The Permo-Triassic extinction. Nature , 367, 231-236. ezaki, y. 1989. Morphological and phylogenetic characteristics of Late Permian rugose corals in Iran. Memoir of the Association of Australasian Palaeontologists , 8. 275-281. — 1993. The last representatives of Rugosa in Abadeh and Julfa, Iran: survival and extinction. Courier Forschungsinstitut Senckenberg , 164. 75-80. — 1994. Patterns and paleoenvironmental implications of end-Permian extinction of Rugosa in South China. Palaeogeography , Palaeoclimatology , Palaeoecology, 107, 165-177. — 1995. The development of reefs across the end-Permian extinction. Journal of the Geological Society of Japan , 101, 857-865. flugel, h. w. 1976. Numidiaphyllidae-eine neue Familie der Rugosa aus dem Ober-Perm von Siid-Tunis. Neues Jahrbuch fur Geologie und Palaontologie , Monatshefte , 9, 54-64. frech, f. 1890. Die Korallenfauna der Trias. I. Die Korallen der juvavischen Triasprovinz. Palaeontographica , 37, 1-116. hallam, a. 1991. Why was there a delayed radiation after the end-Paleozoic extinctions? Historical Biologv, 5, 257-262. hatschek, B. 1888-1891. Lehrbuch der Zoologie, eine morphologische Ubersicht des Thierreiches zur Einfurung in das Stadium dieser Wissenschaft , Lief 1-3. Gustav Fischer, Jena, iv + 432 pp. hill, D. 1981 . Rugosa and Tabulata. F1-F762. In teichert, c. (ed.). Treatise on invertebrate paleontology. Part F. Coelenterata. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 762 pp. iljina, T. G. 1965. [Late Permian and Early Triassic tetracorals from the Transcaucasus area]. Transactions of the Paleontological Institute, USSR Academy of Science, 107, 1-104. [In Russian], — 1984. [Historical development of corals]. Transactions of the Paleontological Institute, USSR Academy of Science, 198, 1-184. [In Russian], Newell, n. d., rigby, j. K., driggs, a., boyd, d. w. and stehli, F. G. 1976. Permian reef complex, Tunisia. Brigham Young University Geology Studies, 23, 75-1 12. nudds, j. r. and sepkoski, j. j. 1993. Coelenterata. 101-124. In benton, m. j. (ed.). The fossil record 2. Chapman and Hall, London, 845 pp. Oliver, w. a. jr 1980. The relationship of the scleractinian corals to the rugose corals. Paleobiology, 6, 146-160. railsback, l. b. and anderson, T. F. 1987. Control of Triassic seawater chemistry and temperature on the evolution of post-Paleozoic aragonite-secreting faunas. Geology , 15, 1002-1005. roniewicz, e. and morycowa, e. 1993. Evolution of the Scleractinia in the light of microstructural data. Courier Forschungsinstitut Senckenberg, 164, 233-240. sando, w. j. 1993. Late Paleozoic coral research: past, present, and future. Courier Forschungsinstitut Senckenberg , 164, 21-36. 14 PALAEONTOLOGY, VOLUME 40 schindewolf, o, h. 1942. Zur Kenntnis der Polycoelien und Plerophyllen. Eine Studie fiber den Bau der “Tetrakorallen” und ihre Beziehungen zu den Madreporarien. Abhandlungen des Reichsamts fur Bodenforschung, Neue Folge , 204, 1-324. scrutton, c. t. 1993. New Kilbuchophyllid corals from the Ordovician of the Southern Uplands, Scotland. Courier Forschungsinstitut Senckenberg, 164, 153-158. — and clarkson, e. n. k. 1991. A new scleractinian-like coral from the Ordovician of the Southern Uplands, Scotland. Palaeontology , 34, 179-194. senowbari-daryan, b. and rigby, j. k. 1988. Upper Permian segmented sponges from Djebel Tebaga, Tunisia. Facies , 19, 171-250. sepkoski, j. j. Jr 1989. Periodicity in extinction and the problem of catastrophism in the history of life. Journal of the Geological Society , London , 146, 7-19. Stanley, G. d. Jr 1988. The history of Early Mesozoic reef communities: a three-step process. Palaios, 3, 170-183. — and swart, p. k. 1995. Evolution of the coral-zooxanthellae symbiosis during the Triassic: a geochemical approach. Paleobiology , 21, 179-199. stevens, c. H. 1975. New Permian Waagenophyllidae (rugose corals) from North Africa. Journal of Paleontology , 49, 706-709. termier, H., termier, G. and vachard, D. 1977. Monographic Paleontologique des affleurements permiens du Djebel Tebaga (Sud Tunisien). Palaeontographica , Abteilung A , 156, 1-109. tucker, m. E. and wright, v. p. 1990. Carbonate sediment ology. Blackwell Scientific, Oxford, 482 pp. wells, j. w. 1956. Scleractinia. F328-F444. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part F. Coelenterata. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 498 pp. wendt, j. 1977. Aragonite in Permian reefs. Nature , 267, 335-337. 1 990c/. The first aragonitic rugose coral. Journal of Paleontology , 64, 335-340. — 19906. Corals and coralline sponges. 45-66. In carter, j. g. (ed.). Skeletal biomineralization : patterns, processes and evolutionary trends , Volume 1. Van Nostrand Reinhold, New York, vii + 832 pp. weyer, d. 1972. Zur Morphologie der Rugosa (Pterocorallia). Geologie, 21, 710-737. yu jianzhang, lin yingdang, shi yen, huang zhuxi and yu xueguang 1983. [Carboniferous and Permian corals], Jilin People’s Publishing House, Jilin, 357 pp. [In Chinese]. YOICHI EZAKI Department of Geological Sciences University of Durham South Road, Durham DH1 3LE, UK Typescript received 24 July 1995 Revised typescript received 26 July 1996 Present address: Department of Geosciences Osaka City University Sugimoto, Osaka 558, Japan UPPER ORDOVICIAN CONODONTS FROM THE KALKBANK LIMESTONE OF THURINGIA, GERMANY by ANNALISA FERRETTI and CHRISTOPHER R. BARNES Abstract. The Kalkbank limestone of Thuringia, Germany, occurs within a condensed clastic-oolitic sequence lying below deposits associated with the Hirnantian (Ashgill) glaciation. Conodonts were first reported from this unit by Kniipfer in 1967 but described only as morphospecies. On the basis of a new collection of over 25000 specimens, 13 species belonging to 11 genera are described. The discovery of M elements of Amorphognathus ordovicicus and A. ventilatus sp. nov. indicates an early Ashgill age for this limestone. Sagittodontina robusta , Scabbardella altipes and Istorinus erectus are the most abundant species; Hamarodus europaeus is also well represented. Taxonomic revisions are made for the first two species. The conodont fauna has a Mediterranean Province affinity and shows close relations with those from the Ashgill of Spain, France and Libya and, to a lesser extent, of the Carnic Alps and Sardinia. Thuringia is one of the northernmost areas of Europe considered to have been part of the cold- water Mediterranean Province in the late Ordovician (Haminann et al. 1982; Sweet and Bergstrom 1984). This province has long been recognized for several different fossil groups, such as trilobites (Whittington and Hughes 1972; Cocks and Fortey 1988), brachiopods (Havlicek 1976), acritarchs (Vavrdova 1974; Martin 1982), chitinozoans (Paris 1981) and conodonts (Sweet and Bergstrom 1984). All the regions belonging to it were located at high latitudes in the southern hemisphere; sedimentation during almost the entire Ordovician was of shallow-water elastics. This depositional pattern was interrupted only at the very end of the period by a single significant interval of carbonate deposition of variable but typically limited thickness regionally. The carbonates are primarily cystoid wackestones that are overlain by a thicker clastic sequence bearing glaciomarine diamictites associated with the Hirnantian (late Ashgill) glaciation (Robardet and Dore 1988; Robardet et al. 1990; Storch 1990). Kniipfer (1967) described the conodont fauna from this carbonate unit in Thuringia, but used only form taxonomy, typical of that time. This monographic work, together with the parallel study of the Tonflaserkalk from the Carnic Alps (Serpagli and Greco 1965; Serpagli 1967), initiated Mediterranean Province conodont investigations. Similar later studies include those of the Urbana Limestone of eastern and western Sierra Morena (Fuganti and Serpagli 1968; Hafenrichter 1979; Sarmiento 1990; Ferretti 1992), the Cystoid Limestone of the eastern Iberian Chain (Carls 1975; Hafenrichter 1979; Ferretti 1992), from Catalonia (Ferretti 1992), the Central Pyrenees in Spain (Hartevelt 1970), the Calcaire de Rosan of Brittany (Lindstrom and Pelhate 1971 ; Paris et al. 1982), the Calcaire de Vaux and calcareous boulders in the Tillite de Feuguerolles of Normandy (Weyant et al. 1977), Austrian Carnic Alps and Northern Greywacke Zone (Schonlaub 1969, 1979, 1980; Flajs and Schonlaub 1976), Sardinia (Helmcke and Koch 1974; Ferretti and Serpagli 1991), Libya (Bergstrom and Massa 1979, 1987, 1992) and Bohemia (Ferretti 1992). An important synthesis of the upper Ordovician European conodont faunas and their biostratigraphy was given by Sweet and Bergstrom (1984), who reconstructed the main multielement taxa of the Thuringian fauna using published descriptions and illustrations. They revised Kniipfer’s identifications at the multielement species level and new data were added later by Dzik (1989) and Fuchs (1989, 1990). Nevertheless, the precise age of the Kalkbank in Thuringia has never been IPalaeontologv, Vol. 40, Part 1, 1997, pp. 15-42, 5 pls| © The Palaeontological Association 16 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Geological map of Schmiedefeld area, Thuringia, after Knupfer (1967), showing sample location. established, as the M element of the Amorphognathus species, which is the critical one for species differentiation in this genus, was missing. A sample which yielded over 25000 conodont elements, collected near Schmiedefeld (Text-fig. 1) and kindly provided by Wolfgang Hammann, forms the basis of this paper, in which the taxonomy of this new fauna and its biostratigraphical and palaeobiogeographical significance are assessed. All the specimens described and figured are housed in the Micropalaeontological Collections of the Department of Earth Sciences of Modena University, Italy (repository numbers IPUM 24721-24866). STRATIGRAPHICAL FRAMEWORK Exposures near Wittmannsgereuth and Schmiedefeld (Text-fig. 1) have provided the best localities in Thuringia for the thin late Ordovician carbonate unit (Kalkbank). A chamosite-rich oolitic level, at the base of the limestone (Text-fig. 2) has supported several small iron mines, now mostly abandoned. In particular, the famous Gebersdorf mine was the source of most of Kniipfer’s samples. Our material comes from near the slag heap of Gebersdorf mine, on the northern slope of the valley about 1 km north-east of Schmiedefeld (Text-fig. 1). The Ordovician of Thuringia is represented by up to 3000 m of terrigenous elastics: in ascending order, on a Cambrian basement, the Frauenbach Group (up to 650 m), the Phycodes Group (up to 2000 m), and the relatively thin Grafenthal Group (about 400 m) which comprises the Griffelschiefer, Schmiedefeld and Lederschiefer formations (Text-fig. 2) of upper Arenig to uppermost Ordovician age. At the base of the Griffelschiefer Formation (40-200 m), consisting of black shales, there is an iron-rich member, the ‘Unterer Erzhorizont’ (0-11 m). The Schmiedefeld Formation, which corresponds to the ’Oberer Erzhorizont’ of Kniifper (1967), is composed of two separate chamositic oolitic levels, the ‘Unteres Erzlager’ (0-2-8 m) and the ’Oberes Erzlager’ (0-20 m), between which a quartzitic unit, the Fagerquarzit, of variable thickness (0-30 m), is usually intercalated. The top FERRETTI AND BARNES: ORDOVICIAN CONODONTS 17 text-fig. 2. Schematic stratigraphical section of the Grafenthal Group, after Kniipfer ( 1967), Steiner and Falk (1981) and Fuchs (1990). Note that the section is not to scale; the condensed section is expanded to show detailed stratigraphical relationships. Thickness of units is : up to 2000 m for the Phycodes Group, 40-200 m for the Griffelschiefer Formation, 0-30 m for the Lagerquarzit, 01-0-4 m for the Kalkbank and 40-250 m for the Lederschiefer Formation. Asterisks indicate iron-rich oolitic horizons. of the formation is formed by the Kalkbank, a light grey, fine grained limestone usually 0- 1-0-4 m thick (Kniipfer 1967), but which may locally be absent (Steiner and Falk 1981). From the Schmiedefeld Formation, conodonts, trilobites, brachiopods, ostracodes, foraminiferans, crinoids, bryozoans and gastropods have been reported (Kniipfer 1967). There is a discontinuity within the formation, between the Lagerquarzit and the 'Oberes Erzlager’, whereas the other strata appear to have been deposited continuously (Kniipfer 1967; Fuchs 1990). A thin slate (0 05-0-5 m) with dark flaser structures overlies the limestone (Steiner and Falk 1981 ) and is followed by the glaciomarine Lederschiefer Formation (40-250 m) containing supposedly ice transported, partly fossiliferous pebbles in the upper part. Rare bryozoans, crinoids, brachiopods and palynomorphs do not allow a precise biostratigraphical age assignment. Event stratigraphical considerations, however, suggest a correlation with similar diamictites of Hirnantian age widespread over the peri-Gondwana shelf. The basal beds of the overlying ‘Untere Graptolithenschiefer' contain graptolites indicative of the Llandovery Parakidograptus acuminatus Biozone (Storch 1990). 18 PALAEONTOLOGY, VOLUME 40 table 1. Number of individual conodont elements recovered from the Kalkbank limestone. Species and elements Number Species and elements Number Amorphognathus ordovicicus Icriodella sp. 1 Pa 356 Istorinus erectus 4832 Pb 321 Panderodus cf. gracilis 1 M 17 Sagittodontina robusta Sa 278 Pa 2749 Sb 221 Pb 246 Sc 198 M 459 Sd 88 Sa 1521 Amorphognathus ventilatus sp. nov. Sb 1098 M 85 Sc 454 Amorphognathus ? sp. 816 Sd 1412 Cornuodus longibasis 6 Scabbardella altipes 6668 Dapsi/odus mutatus 175 Walliserodus sp. 1 Drepanoistodusl sp. 719 'ambalodifornT element 375 Hamarodus europaeus ‘carniodiform' element 381 Pa 231 ‘oistodiform’ element 34 Pb 91 Gen. et sp. indet. 1 M 468 Indet. fragment 256 Sa 79 Total 25150 Sb 96 Sc 397 Sd 19 THE KALKBANK CONODONT FAUNA The 21 kg sample processed dissolved readily in a solution of 10 per cent, formic acid, and produced over 25000 conodont elements. Most have encrusted surfaces, many are fragmentary, and elements belonging to different species commonly have been found fused together. Nevertheless, they could be identified at the multielement species level, and many well preserved specimens also occur. The conodont Colour Alteration Index (CAI) (Epstein et al. 1977) of the specimens is 4, indicating burial temperatures in the range of 190-300 °C. The elements belonging to Sagittodontina robusta and Istorinus erectus appear slightly lighter in colour, but this is probably related to their limited wall thickness, and the cusp of some M elements and the apical part of some Pa and Pb elements of Hamarodus europaeus are sometimes paler than the rest of the specimen. In this conodont fauna, 13 multielement species representing 11 genera have been identified, four of which are left in open nomenclature. The list and abundance of the species is given in Table 1 and Text-figure 3. The fauna is dominated by Sagittodontina robusta (32 per cent.), Scabbardella altipes (27 per cent.) and Istorinus erectus (19 per cent.); Hamarodus europaeus (5 per cent.) is also common. Other species, Walliserodus sp., Panderodus cf. gracilis , Cornuodus longibasis or Icriodella sp. are extremely rare. No stratigraphical or palaeoecological reasons appear to explain the paucity in this fauna of the cosmopolitan coniform genus Panderodus. A different hydrodynamic behaviour of these elements is a possible answer. Selective post-mortem transport could have occurred in which only a fraction of the original fauna was deposited in the Kalkbank unit, resulting in the dominance of apparently light elements such as those of 5. robusta. The relative proportion of species compares closely with that reported by Knupfer (1967) (Text- fig. 3). However, a specific evaluation is difficult since the numerical data of Fuchs (1990) for S. robusta in Kniipfer’s collection include elements of I. erectus , which we regard as a separate species. The great dominance of S. altipes reported by Fuchs in his collection is, as he also suggested. FERRETTI AND BARNES: ORDOVICIAN CONODONTS 19 Amorphognathus ordovicicus Amorphognathus ventilatus sp. nov. Amorphognathus ? sp. Comuodus longibasis Dapsilodus mutatus Drepanoistodus ? sp. Hamarodus europaeus Icriodella sp. Istorinus erectus Panderodus cf. gracilis Sagittodontina robusta Scabbardella altipes Walliserodus sp. 'ambalodiform' element 'carniodiform' element 'oistodiform' element Gen. et sp. indet. Indet. fragment o (N m O c c text-fig. 3. Numbers of individual species recovered from the Kalkbank Limestone sample. probably the result of the aggressive sample processing in the laboratory, that could have destroyed thin-walled elements with a deep basal cavity, such as those of S. robusta. In addition to the previously known fauna, species of Icriodella and Comuodus are reported for the first time. After comparison with collections of Ashgill conodonts from Whitland, South Wales, we doubt that ‘ Ambalodus triangularis triangularis Branson and Melil. 1993 ’ from Knupfer’s collection ( 1967, pi. 9, fig. 2a-b) and present in our fauna (PI. 2, figs 11-13) could represent the Pb element of Rhodesognathus as previously supposed (Dzik 1989; Fuchs 1990). In fact the lateral process of the Pb elements of Rhodesognathus develops as an extension of the denticle anterior to the cusp, and not, as happens in our specimens, of the cusp itself. For the same reason, Bergstrom (1983, p. 46) regarded the 'Prioniodus' ( Rhodesognathus ?) n. sp. a IT. Prioniodus variabilis Bergstrom, 1962 and Prioniodus gerdae Bergstrom, 1971 reported by Lindstrom et al. (1974) from Brittany as probable Prioniodus (Baltoniodus) rather than Rhodesognathus. Note that Dzik ( 1994) has recently underlined how the position of branching of the anterior process in the platform elements may be a useful character for distinguishing Rhodesognathus. After direct comparison with the material described from Spain by Carls (1975) and Flafenrichter (1979), their corresponding specimens of ‘ Ambalodus triangularis triangularis Branson and Mehl, 1933’ appear similar to ours. In addition, the elements classified as 1 Ambalodus robustus Rhodes, 1953’ by Kniipfer (1967, pi. 10, figs 7-8) are more likely, in our opinion, to be anterior processes of an ‘amorphognathiform’ platform that we have left in open nomenclature (PI. 2, figs 1-4), and not the Pa elements of Rhodesognathus (Dzik 1989; Fuchs 1990). The specimens of Rhodesognathus elegans reported by Lindstrom and Pelhate (1971) from the Calcaire de Rosan in Brittany and by Sarmiento (1990) from the Urbana Limestone of eastern Sierra Morena were neither figured nor described, and so it is impossible to attempt any comparison, but, to date, represent the only reports of this species in the Mediterranean Province. New investigations will probably reveal whether this form can be regarded as a palaeogeographical marker for lower latitudes and of warmer water compared with the cool water environment inferred for the Kalkbank. THE AGE OF THE KALKBANK LIMESTONE Kniipfer (1967) described his fauna using form element taxonomy; 58 species belonging to 26 genera were recognized, including the important new genera Sagittodontina and Istorinus. On the basis of conodont and trilobite faunas, Kniipfer suggested that the age of the limestone was close to the Caradoc-Ashgill boundary. 20 PALAEONTOLOGY, VOLUME 40 Subsequently, in a general study of Amorphognathus , Lindstrom (1977) referred the morpho- species figured by Knupfer to A. ordovicicus. Sweet and Bergstrom (1984) reinterpreted the Thuringian fauna and recognized nine multielement species including A. cf. ordovicicus. The identifications in both these studies were based mainly on the platform elements figured by Knupfer, as no unequivocal M elements were recognized. In a different approach, Dzik (1989) recognized A. superbus , based on the assumption that the single specimen of ‘ Holodontus sp.’ in Kniipfer’s collection (pp. 30-31, pi. 3, fig. 14) was the M element of the species. In addition, he also observed that Pa elements of A. superbus are generally larger than the corresponding ones of A. ordovicicus. Fuchs (1990), by the use of acetic and phosphoric acid, recovered about 250 additional elements, but no M elements. Nevertheless, he assigned the Kalkbank to the A. superbus Zone (Caradoc to lower Ashgill), basing his identification on the specimen of ‘ Holodontus sp. ’ mentioned above. The presence of many M elements in our fauna represents an important discovery. Seventeen of A. ordovicicus have been found together with 85 of A. ventilatus sp. nov., and are described below. The latter lack the anterior aboral denticle that is typical of ‘ Holodontus superbus Rhodes, 1953’, the M element of Amorphognathus superbus. However, they do possess a very long anterior aboral process and the cusp is not discrete as in ‘ Goniodontus superbus Ethington, 1959’, the corresponding M element of A. ordovicicus. On the outer lateral edge of the cusp, and not on the outer lateral process, they generally have two or three small denticles growing at a different angle towards the axis of the cusp. All these features are more typical of the M element in A. superbus , although some features of the A. ordovicicus M element are definitely present. It is difficult to judge from a single sample, even with an abundant fauna, whether this new species represents a transitional evolutionary stage. A similar form was figured by Viira (1974, p. 90, fig. 110) from the Ohesaare core from Estonia, but unfortunately no other elements of that fauna indicate taxonomic or stratigraphical relationship. The horizon from which her element was recovered is immediately below an ordovicicus fauna (Lindstrom 1977), while a superbus fauna occurs in the lowermost part of the core interval referred to the Nabala Stage (Bergstrom 1971). Lindstrom (1977) regarded this element, with question, as the M element of A. complicatus , but in the Kalkbank material no platform elements of this latter species occur to verify this possibility. The different elements found could presumably represent a reworked assemblage. However, there are no significant differences in preservation that are typical for reworked conodonts. This potential explanation is rejected. Nevertheless, the evolution of the M element in the Amorphognathus lineage, starting from the older A. tvaerensis, shows a gradual reduction in number, and final disappearance of the denticles on the inner anterior process, a contraction of the anterior aboral process which eventually disappears in A. ordovicicus , the development of a prominent and discrete cusp, a reduction of its inclination, and an increase in cusp size. It is noteworthy that Orchard (1980) reported some M elements, lacking accessory denticles on the cusp but with three well-developed processes, as A. aff. superbus , from the upper part of the superbus zone from the Crug and Birdshill limestones in South Wales. In summary, we consider the age of the Kalkbank limestone to be early Ashgill, equivalent to the base of the ordovicicus Zone. The similar limestones that were deposited in other parts of the Mediterranean Province appear to be slightly younger, but within this zone (Ferretti 1992). This difference may be in response to ecological and/or oceanographical events that initiated calcareous sedimentation diachronously in the diverse areas. Finally, only the discovery of additional abundant samples, perhaps in other areas of Thuringia, and new data will reveal whether the Kalkbank represents a condensed unit equivalent to entire sequences deposited during the Ashgill elsewhere in southern Europe, or only to their older part. TAXONOMIC REMARKS: THE AMORPHOGNATHUS LINEAGE Although the Kalkbank fauna was recovered from a single sample and the faunal composition is similar in general to earlier reports (Knupfer 1967; Dzik 1989; Fuchs 1989, 1990), it is its sheer FERRETTI AND BARNES: ORDOVICIAN CONODONTS 21 abundance that is remarkable, and which allows a more complete taxonomic description of the dominant species. In particular, there is a substantial number of specimens of three forms of Amorphognathus (A. ordovicicus , A. ventilatus sp. nov.. A.? sp. ). This study, together with one nearing completion on coeval basal Ashgill faunas from Whitland, South Wales raises doubts concerning the rather simple pattern of evolution in the Amorphognathus lineage previously proposed (e.g. Bergstrom 1971, 1983; Sweet and Bergstrom 1976; Dzik 1989, 1994), i.e. from A. tvaerensis through A. superbus to A. ordovicicus on which the upper Ordovician conodont biozonation is based. The relationship of other late Ordovician species of Amorphognathus (e.g. A. complicatus,A. lindstroemi) to this main lineage has not been established ; these two species are less widely distributed and perhaps overlooked. As noted above and described below, the M element of A. ventilatus sp. nov. bears some peculiar morphological characteristics that suggest that it may have evolved from A. superbus. In addition to the tvaerensis-superbus-ordovicicus lineage there may be at least one other in the late Ordovician, culminating in A. lindstroemi. As partly reflected in our synonymies for Amorphognathus species, we suspect that the complexity of evolution within Amorphognathus has been masked by some earlier workers assigning different forms into one or more of the three species in the main lineage. With new studies under way it may be possible to refine further the upper Ordovician conodont biozonation through an improved understanding of phylogenetic relationships within Amorphognathus. Aldridge et al. (1995) established the apparatus architecture of Promissum pulchrum Kovacs- Endrody based on over 100 complete apparatuses preserved on bedding planes of the upper Ordovician Soom Shale Member, Cedarberg Formation, South Africa. Promissum , like Amorphognathus , has a prioniodontid apparatus plan, but includes not only Pa and Pb elements, but also Pc and Pd elements. Aldridge et al. (1995) considered that other Ordovician prioniodontid taxa may also include Pc and/or Pd elements (e.g. Gamachignathus). From the Kalkbank material, we have yet to recognize elements in Amorphognathus that could be assigned unequivocally as Pc or Pd elements. PALAEOGEOGRAPHICAL REMARKS Conodonts belonging to the Mediterranean Province have a special value, as they represent a high latitude fauna living in polar to subpolar environments (Sweet and Bergstrom 1984; Bergstrom and Massa 1992). Some of them are typical of this region, and are absent from coeval faunas of inferred lower and middle latitudes. Sagittodontina robusta and Istorinus erectus are probably the best indicators of the Mediterranean Province. So far, both of these species seem to be most abundant in Thuringia, but this may reflect a slight age difference and/or ecological factors. This Kalkbank fauna reinforces the strong faunal similarity noted above between Thuringia, Libya, Spain and France. Elements of S. cf. robusta have been found also in the Pernik Bed of Bohemia (Ferretti 1992), but the fauna is poor and too sparse to allow more specific conclusions. The two indicator species have so far not been found in Sardinia and the Carnic Alps. This latter region appears to have the most diversified fauna in southern Europe and to have had closer relations to more temperate faunas, such as those in Britain, having several taxa in common that have not been reported elsewhere in southern Europe. Bohemia (Marek and Havlicek 1967; Havlicek 1976, 1977), Sardinia (Leone et al. 1991) and the Carnic Alps (Schonlaub 1971 ; Jaeger et al. 1975) are so far the only regions in southern Europe from which a shelly Hirnantian fauna has been reported. This probably reflects different latitudinal/ecological conditions. Finally, Rhodesognathus elegans could prove to be a valuable index species of more temperate regions. From South Africa, another area of Gondwanaland within high latitudes in the late Ordovician, a new conodont fauna has been described (Theron et al. 1990; Aldridge et al. 1994; Aldridge et al. 1995). Over 100 bedding plane assemblages of Promissum pulchrum have been found, some preserving soft tissue impressions. The elements are up to an order of magnitude larger than most late Ordovician conodonts. The palaeobiogeographical and palaeoecological significance of this fauna is not yet clear. The monospecific fauna occurs in a subpolar, non-carbonate facies and has 22 PALAEONTOLOGY, VOLUME 40 not been found elsewhere. It could represent an even colder water fauna than that of the Mediterranean Province. Only recently has more attention been directed towards the thin iron-rich oolitic horizons below the limestones that are a feature typical of the different sequences in south-west Europe and adjacent areas, including the Ossa Morena Zone, Celtiberia, Normandy, Brittany, Libya, Thuringia and Bohemia (Young 1992, fig. 1). Young (1989) recognized these horizons at three stratigraphical intervals in the Ordovician: lower Llanvirn, the lower Caradoc and the lower Ashgill. They represent ‘sediments formed under extremely low sedimentation rates and were most commonly developed as the initial deposit above a disconformity ... [they] form important marker horizons for correlation within the region’ (Young 1992, p. 321). We consider it probable, therefore, that iron-rich oolitic levels were deposited in the early Ashgill almost simultaneously over most of the Gondwana shelf and initiated a new depositional sequence with basal terrigenous deposits, overlain by younger limestones in some shallower and/or higher latitude areas (i.e. Celtiberia or Ossa Morena) and already calcareous in other apparently more offshore and/or lower latitude and possibly condensed sequences, as in Thuringia. SYSTEMATIC PALAEONTOLOGY Orders and families are mostly from Sweet (1988). Synonymy lists are limited to references to the first description of morphospecies, first apparatus reconstruction and recent reports. Descriptions are given only if new information is provided by the fauna examined. Diagnoses are limited to new species. For complete data, see Dzik (1994). Order belodellida Sweet, 1988? Family ansellidae Fahraeus and Hunter, 1985? Genus hamarodus Viira, 1974 Type species. Distomodus europaeus Serpagli, 1967. Hamarodus europaeus (Serpagli, 1967) Plate 3, figures 1-14 1955 Microcoelodusl sp. Rhodes, p. 133, pi. 10, figs 15, 19, 22. 1955 Cordylodus elongatus Rhodes; Rhodes, p. 135, pi. 7, figs 5-6. 1959 Oistodus n. sp. Lindstrom, p. 440, pi. 3, fig. 13. 1959 Cordylodus n. sp. Lindstrom, p. 438, pi. 3, figs 34-36. 1964 INeoprioniodus brevirameus Walliser, p. 47, pi. 4, fig. 5; pi. 29, figs 5-10. 1964 IRoundya prima Walliser, p. 71, pi. 4, fig. 6; pi. 31, figs 1-2. 1966 Oistodus breviconus Branson and Mehl; Hamar, p. 63, pi. 1, fig. 19; text-fig. 4 (11). 1966 N. genus and n. sp. Hamar, p. 77, pi. 3, figs 8-10; text-fig. 5 (5a-b). 1967 Distomodus europaeus Serpagli, p. 64, pi. 14, figs 1-6. 1967 ‘ Oistodus ’ niger Serpagli, p. 79, pi. 20, figs 1-7. 1967 Oistodus abundans Branson and Mehl; Kniipfer, p. 34, pi. 5, fig. 4. 1976 Hamarodus europaeus (Serpagli); Dzik, p. 435, text-fig. 36. 1980 Hamarodus europaeus (Serpagli); Orchard, p. 21, pi. 4, figs 22, 25, 29-31. 1985 Hamarodus europaeus (Serpagli); Bergstrom and Orchard, pi. 2.5, figs 4, 7, 12. 1989 Hamarodus europaeus (Serpagli); Dzik, text-fig. 16. 1991 Hamarodus europaeus (Serpagli); Ferretti and Serpagli, pi. 2, figs 1-6. 1994 Hamarodus brevirameus (Walliser); Dzik, p. Ill, pi. 24, figs 14—19; text-fig. 31a. Remarks. Hamarodus europaeus is common in our fauna, in which all the morphotypes are present. Surprisingly, Kntipfer’s (1967) collection lacks Pa and Pb elements, which are the most typical representatives of the species. In our material, they are laterally compressed with a deep, wide basal FERRETTI AND BARNES: ORDOVICIAN CONODONTS 23 cavity and sharp anterior, upper and posterior margins. The wall is rather thin in most specimens, and, as a consequence, appears lighter in colour. The apical part of the cusp is commonly paler. The terminal denticulation of the anterior and upper keels is commonly broken. Pa elements are more abundant than Pb elements. The M element is the most common component of the apparatus. It displays high variability in the angle between the posterior and upper margin and in the profile of the anterior margin, which may be variably curved. Lateral faces have no costae or other ornamentation. Only the larger elements appear completely black, while the smaller ones have a paler colour. The ratio of P to M elements in our fauna is 0-69, closer to that in Serpagli’s (1967) material from the Carnic Alps (046) than to that in Orchard’s (1980) collection from Great Britain (0-29). The posterior process of S elements, extending at about 90° from the cusp, has a hindeodellid denticulation pattern. Fragments of these bars were probably regarded by earlier authors as ‘carniodiform’ elements. Distinction between the Sb and Sd elements is difficult, due in part to the poor preservation of our material. As the anterior denticles are not always recognizable, the presence of the lateral ones has been considered diagnostic. We follow Sweet (1988) in placing Hamarodus questionably in the family Ansellidae. Occurrence. Upper middle and upper Ordovician of Europe (Thuringia, Carnic Alps, Spain, ?France, Great Britain, Estonia, Norway, Sardinia, Sweden, Bohemia, Poland) and China. Family belodellidae Khodalevich and Tschernich, 1973 Genus walliserodus Serpagli, 1967 Type species. Acodus curvatus Branson and Branson, 1947. Wcdliserodus sp. Plate 1, figure 24 Description. Only one fragment, lacking the basal area, is present. Two prominent costae are developed on the lateral faces of the element but they fade apically. Cross section outline slightly biconvex. Occurrence. Walliserodus is known from the upper Ordovician and Silurian of Europe, North America, Australia and Asia. Family dapsilodontidae Sweet, 1988 Genus dapsilodus Cooper, 1976 Type species. Distacodus obliquicostatus Branson and Mehl, 1933. Dapsilodus mutatus (Branson and Mehl, 1933) Plate 3, figures 15-19 1933 Belodusl mutatus Branson and Mehl, p. 126, pi. 10, fig. 17. 1959 Acodus inornatus Ethington, p. 268, pi. 39, fig. 11. 1959 Distacodus procerus Ethington, p. 275, pi. 39, fig. 8. 1967 Acodus curvatus Branson and Branson; Serpagli, p. 41, pi. 6, fig. 3a-c. 1967 Acodus mutatus (Branson and Mehl); Serpagli, p. 41, pi. 6, figs la-b, 6a-b. 1967 Acontiodus procerus (Ethington); Serpagli, p. 46, pi. 9, figs 6-11. 1980 Dapsilodus mutatus (Branson and Mehl); Orchard, p. 20, pi. 5, figs 6, 15-16, 21. 1994 Dapsilodus mutatus (Branson and Mehl); Dzik, p. 64, pi. 11, figs 24-26, 31-35; pi. 14, figs 8-9; text-fig. 6d. 24 PALAEONTOLOGY, VOLUME 40 Remarks. Representatives of Dapsilodus mutatus are relatively common in our collection. Little can be added to the detailed observations already given by earlier authors on the single morphotypes. In addition, the poor preservation prevents the recognition of the antero-aboral costae on the lateral faces described in the Keisley Limestone specimens by Orchard (1980) or the striations figured in the Swedish specimens by Lofgren (1978). The morphospecies ‘ Oistodus venustus Stauffer, 1935’ is absent in our fauna; another type of ‘ oistodiform’ element of uncertain taxonomic status has been left in open nomenclature. Occurrence. Middle and upper Ordovician of Europe (Great Britain, Thuringia, Sardinia, Carnic Alps, Spain, Bohemia, Norway, Sweden, Estonia, Poland), North America, China and Libya. Order panderodontida Sweet, 1988 Family panderodontidae Lindstrom, 1970 Genus panderodus Ethington, 1959 Type species. Paltodus unicostatus Branson and Mehl, 1933. Panderodus cf. gracilis (Branson and Mehl, 1933) Plate 1, figure 23 cf. 1933 Paltodus gracilis Branson and Mehl, p. 108, pi. 8, figs 20-21. cf. 1967 Panderodus gracilis (Branson and Mehl); Serpagli, p. 85, pi. 23, figs 3-5. cf. 1976 Panderodus gracilis (Branson and Mehl); Dzik, p. 428, fig. 15a-b, e-f. Remarks. Only one small but complete specimen is present in our collection. A principal costa is developed on each lateral face of the element; minor longitudinal striations run close to the upper margin. The single specimen does not allow an assignment within the new apparatus architecture model which was proposed recently for Panderodus by Sansom et al. (1995). Occurrence. P. gracilis is well represented in the Ashgill of Europe, North America and Asia and ranges from the middle Ordovician into the Silurian. explanation of plate 1 Figs 1-15. Amorphognathus ordovicicus Branson and Mehl, 1933. 1-3, IPUM 24721, IPUM 24722, IPUM 24723; upper views of Pa elements; x 130, x 95 and x 95. 4-5, IPUM 24724, IPUM 24725; antero-lateral views of sinistralPb elements; x 130 and x 95. 6-9, IPUM 24726, IPUM 24727, IPUM 24728, IPUM 24729; postero-lateral views of M elements; xllO, x 130, xllO, x 100. 10-11, IPUM 24730, IPUM 24731; antero-lateral views of Sc elements; x 120, x 105. 12, IPUM 24732; lateral view of Sa element; x 120. 13, IPUM 24733; lateral view of Sd element; x 120. 14-15, IPUM 24734, IPUM 24735; lateral views of Sb elements; x 125, xllO. Fig. 16. Icriodella sp.; IPUM 24736; upper view of anterior process of Pa element; x 50. Figs 17-22. Scabbardella altipes (Henningsmoen, 1948). 17, IPUM 24737; x60; 18, IPUM 24738; x60; 19, IPUM 24739; x60; 20, IPUM 24740; x45; 21, IPUM 24741; x60; 22, IPUM 24742; x45. All lateral views. Fig. 23. Panderodus cf. gracilis (Branson and Mehl, 1933); IPUM 24743; lateral view; x 100. Fig. 24. Walliserodus sp.; IPUM 24744; lateral view; x95. PLATE 1 FERRETTI and BARNES, Kalkbank conodonts 26 PALAEONTOLOGY, VOLUME 40 Order prioniodontida Dzik, 1976 Family balognathidae Hass, 1959 Genus amorphognathus Branson and Mehl, 1933 Type species. Amorphognathus ordovicica Branson and Mehl, 1933. Amorphognathus ordovicicus Branson and Mehl, 1933 Plate 1, figures 1-15 1933 Amorphognathus ordovicica Branson and Mehl, p. 127, pi. 10, fig. 38. 1933 Ambalodus triangularis Branson and Mehl, p. 128, pi. 10, figs 35-37. 1955 Roundya diminuta Rhodes, p. 137, pi. 8, figs 9, 12; pi. 9, fig. 6. 1955 Keislognathus gracilis Rhodes, p. 131, pi. 7, figs 7-8; pi. 8, figs 10, 13-16. 1955 Ligonodina cf. L. elongata Rhodes; Rhodes, p. 134, pi. 8, figs 7-8. 1955 Rosagnathus superbus Rhodes, p. 129, pi. 7, figs 1-4. 1959 Goniodontus superbus Ethington, p. 278, pi. 40, figs 1-2. 1971 Amorphognathus ordovicicus Branson and Mehl; Bergstrom, p. 134, pi. 2, figs 6-7. 1978 Amorphognathus ordovicicus Branson and Mehl; Bergstrom, pi. 80, figs 1-11. 1980 Amorphognathus ordovicicus Branson and Mehl; Orchard, p. 16, pi. 4, figs 1-13, 17-18. 1983 Amorphognathus ordovicicus Branson and Mehl; Nowlan, p. 660, pi. 2, figs 16-17, 22, 25-27. 1986 Amorphognathus ordovicicus Branson and Mehl; Savage and Bassett, p. 691, pi. 84, figs 1-21; pi. 85, figs 1-26; pi. 86, figs 1-13. 1991 Amorphognathus ordovicicus Branson and Mehl; Ferretti and Serpagli, pi. I, figs 1-9. 1992 Amorphognathus sp. cf. A. ordovicicus Branson and Mehl; Bergstrom and Massa, p. 1337, pi. 1, figs 18-24. 71994 Amorphognathus ordovicicus Branson and Mehl; Dzik, p. 94, pi. 23, figs 6-12; ?pl. 24, fig. 20; text-figs 21c, 22. Remarks. No intact Pa elements are present in our fauna, and only a few are complete enough to allow a specific identification. Nevertheless, both sinistral and dextral forms have been found, and the blade type and non blade type of anterior process described by Bergstrom (1964) are present. Only isolated or coupled processes have been counted in the attempt to give a more realistic numerical estimation of the fauna. Pb elements display high variability, both in sinistral and dextral form, involving the general size, the angle between the anterior and posterior processes and the upper profile of the anterior process. Some larger specimens have the anterior and posterior processes more aligned; in some smaller elements the upper margin of the anterior process appears more recurved in outer lateral view. The distinction between these two types is not always clear, as many transitional forms are present. EXPLANATION OF PLATE 2 Figs 1-10. Amorphognathusl sp. 1-4, upper view and lateral views of IPUM 24745; x 100; IPUM 2746; x 100; IPUM 24747; x90; IPUM 24748; x95. 5-9, upper views of IPUM 24749; x 100; IPUM 24750; x90; IPUM 24751 ; x 90; IPUM 24752; x 125; IPUM 24753; x 95. 10, IPUM 24754; lateral view; x 75. Figs 11-13. ‘ambalodiform’ elements. 1 1, IPUM 24755; x 90. 12, IPUM 24756; x95. 13, IPUM 24757; x 90. All lateral views. Figs 14-17. Amorphognathus ventilatus sp. nov. 14, IPUM 24758; antero-lateral view; x 120. 15a-c, holotype, IPUM 24759; antero-lateral, oral and postero-lateral views; x 150, x 170, x 165. 16, IPUM 24760; antero- lateral view; x 150. 17, IPUM 24761; lateral view; x 135. Figs 18-20. Cornuodus longibasis (Findstrom, 1955). 18, IPUM 24762; x90. 19, IPUM 24763; x 130. 20, IPUM 24764; x95. All lateral views. Fig. 21. Gen. et sp. indet.; IPUM 24765; lateral view; x95. PLATE 2 FERRETTI and BARNES, Kalkbank conodonts 28 PALAEONTOLOGY, VOLUME 40 Savage and Bassett (1986) regarded the variability of the Pb element as a possible key to identification of Amorphognathus species, being smaller and more robust in A. ordovicicus compared with the corresponding elements of older species. Recently, Bergstrom and Massa (1992) rejected this interpretation, as the Pb element appeared too similar in the different species of the genus to be diagnostic. It is impossible to establish if the variability observed in our specimens represents different species or simply intraspecific variation. In describing the apparatus of Promissum pulchrum from the upper Ordovician of South Africa, Aldridge et al. (1995) established the presence of Pc and Pd elements in addition to Pa and Pb elements within the prioniodontid apparatus plan. They suggested that other Ordovician prioniodontid genera (e.g. Gamachignathus ) and certainly some Silurian (e.g. Pterospathodus , Coryssognathus , Astropentagnathus) may also possess Pc and/or Pd elements. Despite the high number of platform elements, we are unable to recognize unequivocally Pc or Pd elements within the apparatus of Amorphognathus as represented in the Kalkbank collection. M elements show the typical features described by Ethington (1959) and subsequently redefined by Serpagli (1967). The anterior inner lateral process is reduced in all our specimens. The anterior outer lateral process, if not broken, is always denticulated, carrying a main denticle and some minor ones distally, located in the space between the cusp and the main outer denticle or on the outer process. In one specimen (PI. 1, fig. 7) a small denticle was observed close to the main lateral denticle. The posterior process can be adenticulated, even if long, or may carry a single small denticle. S elements are present in different proportions, being the Sa, Sb and Sc in similar proportions, while the Sd elements are rare. The approximate ratio of Pa, Pb, M and S elements is respectively 21:19:1:46. This value is only approximate as we have preferred to keep separate some fragments of a possible Pa element of Amorphognathus ? sp. and some M elements of A. ventilatus sp. nov., that could have shared many of the elements here included in A. ordovicicus. Occurrence. Amorphognathus ordovicicus is well known in the upper Ordovician of Europe, North America and part of Asia. It is known in Europe from Spain, north-west France, the Carnic Alps and adjacent areas of Austria, Sardinia, Thuringia, ?Bohemia, Poland, Sweden, Estonia and the British Isles. Amorphognathus ventilatus sp. nov. Plate 2, figures 14-17 1974 Holodontus n. sp. Viira, p. 90, fig. 110 (only). Derivation of name. From the Latin ventilare (= to fan). Holotype. M element IPUM 24759; PI. 2, fig. 15a-c; Kalkbank limestone, Gebersdorf mine, Schmiedefeld, Thuringia, Germany. Diagnosis. Compound M ('holodontiform') element consisting of a cusp from which three processes develop. General profile that of a deformed pyramid, with unequal triangular faces, each process representing an edge. Inner anterior process is thin, long and has no denticles. On outer edge of cusp, at mid-height, generally two or three sharp denticles are developed with a divergent direction of growth to axis of cusp. Denticles commonly fused together in a kind of fan. Description. Both sinistral and dextral forms are present. The oral margin appears quite asymmetrical, being completely straight on the inner side, and denticulated on the outer. Lateral denticles are generally smaller than the cusp, which has a prominent position on element. The posterior process is directly connected to the anterior outer process with a ridge, giving a typical triangular profile in outer-lateral view. The anterior face of the cusp is planar to slightly convex; the posterior face is convex. The cusp is slightly flexed postero-laterally. The posterior process is non-denticulated. In some specimens the oral edge of the posterior process extends to the FERRETTI AND BARNES: ORDOVICIAN CONODONTS 29 outer edge of fan, and not along the cusp. The large basal cavity is triangular in aboral view and extends beneath the processes as a groove. Remarks. Only the M element of A. ventilatus sp. nov. has been clearly recognized. Other elements of the apparatus are probably conflated with those of A. ordovicicus and possibly, with Amorphognathusl sp. A specimen similar to our M elements was described and figured by Viira (1974, p. 90, fig. 1 10) as ‘ Holodontus n. sp.’ from the Ohesaare core in Estonia (depth 461-95 m). Three elements from the overlying formation (Viira 1974, pi. 13, figs 29-31) were included in the same morphospecies, but probably belong to A. ordovicicus (see Lindstrom 1977). Viira’s (1974, fig. 110) ‘ Holodontus n. sp.’ appears to have a small denticle on the inner lateral process, a feature not observed in our specimens. Lindstrom (1977) included this element questionably in Amorphognathus complicatus. Unfor- tunately, we have few platform elements complete enough to allow specific identification, but none belongs to that species. Some of the ‘ambalodiform’ elements of our fauna resemble ‘ Ambalodus frognoeyensis Hamar, 1966’, the Pb element of A. complicatus , but we do not regard it as diagnostic. The single specimen of ‘ Holodontus sp.’ described by Kniipfer (1967) undoubtedly shows, according to his description and illustration, many similarities to the M element of A. ventilatus sp. nov., but it appears to lack a real cusp. Two M elements of A. superbus described by Savage and Bassett (1986, pi. 83, figs 13-14) seem to develop two denticles only on one edge of a main denticle, which is not sufficiently prominent to be considered a real cusp. The characteristic disposition of the denticles in a kind of fan on the outer side of the cusp seems typical of the species and has not been recorded in other ‘holodontiform’ elements. In addition, the M element of A. ventilatus sp. nov. differs from ’//. superbus Rhodes, 1953’ in the absence of the anterior aboral denticle, and from ‘ Goniodontus superbus Ethington, 1959’ in retaining a well- developed anterior process. A similarity, especially in outer-lateral view, occurs in some specimens of ‘G. lindstroemi Serpagli, 1967’ (pi. 16, figs 2^1), but the latter lack a long anterior inner process and the ray of denticles on the outer edge of the cusp. The possible phylogenetic relationships of this element to other species of Amorphognathus are discussed above. Occurrence. Upper Ordovician of Thuringia and Estonia. Amorphognathus ? sp. Plate 2, figures 1—10 1967 Ambalodus robustus Rhodes; Kniipfer, p. 20, pi. 10, figs 7-8, ?9. 1990 Rhodesognathus elegans (Rhodes); Fuchs, p. 206, pi. 6, fig. 3. Description. Several Pa fragments present in our fauna are described separately because of: their larger size compared with platform elements of A. ordovicicus ; the special denticulation of platform; the dark colour; and presence of a prominent cusp on the anterior process. Two variants are recognized. One type (Kniipfer 1967, p. 20, pi. 10, figs 7-8) is represented by narrow blade-shaped processes (PI. 2, figs 1-4). In oral view, they are straight or slightly sinuous, with submedian crest with rather high and narrow denticles that generally increase in size towards the cusp. Only one of them, although badly preserved, was complete enough to show, posterior to the cusp, the beginning of a platform bearing two parallel denticles with thin costa developing at 90° from one of them (PI. 2, fig. I ). A well-developed ledge runs along the basal margin and continues along the lateral process. On one side of the cusp, a shallow, narrow groove continues directly into what was probably the lateral process, but this is never preserved completely in our specimens. The cusp may be erect or slightly recurved posteriorly and has rounded edges. Between the cusp and platform is an area that is undenticulated or which bears one or two small denticles. The outer margin is generally straight, and then opens on a platform. In lateral view of very sinuous forms, the outer basal edge of process may be visible. The basal cavity is narrow and deep. A second type (Kniipfer 1967, ?pl. 10, fig. 9) is represented by wider fragments, commonly arched, that may be non-blade anterior processes (PI. 2, fig. 1 0). The longitudinal median crest is composed of smaller, wider and equidimensional denticles. The basal ledge is commonly missing. 30 PALAEONTOLOGY, VOLUME 40 Platforms that could have been attached to these processes are never complete, and have a lighter colour. The best preserved (PI. 2, figs 5-9) are composed of both main (posterior) process, probably growing along the same line of cusp, and another bilobate (inner) process. A darker edge is present almost all around the platform. The posterior process has a variable length: it can be of the same size as the posterior-inner one or longer. The anterior-inner bar is at almost 90° from the posterior-inner one. Denticles are nodose and disposed slightly towards the inner side of the process, and are not exactly central. Those of the posterior-inner process run almost parallel to those of the posterior process. The anterior-inner process has maximum of three denticles developed along a line that merges with the first denticle of the posterior-inner process. The posterior process may develop another denticulation just at the outer-posterior extremity, with a maximum of three primitive denticles or nodes (PI. 2, fig. 5). Remarks. If compared with the platforms of Amorphognathus ordovicicus and A. superbus , Amorphognathusl sp. appears primitive both in denticulation and in disposition of the processes, and it is more similar to older species of the genus such as A. inaequalis. The platforms described above could also represent, as suggested by Bergstrom (pers. comm. 1993), the non-blade element of Sagittodontina robusta. All the types reported here are also present in the upper Ordovician of Spain (Ferretti 1992). Occurrence. Upper Ordovician of Thuringia and Spain. Genus sagittodontina Kntipfer, 1967 Type species. Sagittodontina robusta Kntipfer, 1967. Sagittodontina robusta Kniipfer, 1967 Plate 4, figures 1-23 1967 Sagittodontina robusta Kntipfer, p. 38, pi. 8, figs 3a-b, 4. 1967 Sagittodontina separata Kniipfer, p. 38, pi. 8, figs 5, 7a-b. 1967 Sagittodontina unidentata Kniipfer, p. 39, pi. 8, figs la-b, 2a-b. 1967 Sagittodontina bifurcata Kniipfer, p. 39, pi. 7, fig. 5a-b. 1967 Sagittodontus dentatus (Ethington); Kniipfer, p. 37, pi. 7, fig. 6a-b. 71967 Sagittodontus robustus robustus Rhodes; Kniipfer, p. 36, pi. 4, fig. 7a-b; pi. 5, fig. 9a-c. 71967 Sagittodontus robustus flammeus Kniipfer, p. 37, pi. 5, fig. 8a-b; pi. 11, figs 9-10. 1967 Zygognathus latypica Kniipfer, p. 43, pi. 5, figs 5a-b, 7a-b. 71967 Acodus deltatus altior Findstrom; Kniipfer, p. 17, pi. 4, figs 2-6. 1967 Ligonodina ?sp. Kniipfer, p. 33, pi. 4, fig. 14. 1967 Strachanognathus thuringensis Kniipfer, p. 40, pi. 5, fig. 6a-b. 1967 Zygognathus asymmetrica Kniipfer, p. 42, pi. 6, figs la-b, 2a-b. 1967 Trichonodella n. sp. Kniipfer, p. 41, pi. 6, figs 3a-b, 4. EXPLANATION OF PLATE 3 Figs 1-14. Hamarodus europaeus (Serpagli, 1967). 1-2, IPUM 24766; x 80; IPUM 24767; x 90; Pa elements. 3-4, IPUM 24768; x 95; IPUM 24769; x 75; Pb elements. 5-6, IPUM 24770; x 100; IPUM 24771 ; x 100; M elements. 7-8, IPUM 24772; x 90; IPUM 24773; x 100; Sc elements. 9-10, IPUM 24774; x 100; IPUM 24775; x 100; Sa elements. 1 1, IPUM 24776; x 1 15; Sb element. 12-14, IPUM 24777; x 120; IPUM 24778; x 140; IPUM 24779; x 120; Sd elements. All lateral views. Figs 15-19. Dapsilodus mutatus (Branson and Mehl, 1933). 15, IPUM 24780; x 100. 1 6, IPUM 24781 ; x 105. 17, IPUM 24782; x 85. 18, IPUM 24783; x 115. 19, IPUM 24784; x 110. All lateral views. Figs 20-22. ‘carniodiform’ elements. 20, IPUM 24785; x 145. 21, IPUM 24786; x 100. 22, IPUM 24787; x 150. All lateral views. Figs 23-24. ‘oistodiform’ elements. IPUM 24788; x 100. IPUM 24789; x 130. Both lateral views. PLATE 3 FERRETTI and BARNES, Kalkbank conodonts 32 PALAEONTOLOGY, VOLUME 40 1967 Tripodontus muelleri Kniipfer, p. 42, pi. 6, figs 5a -b, 6a-b. 1967 Tripodontus compactus Kniipfer, p. 41, pi. 6, figs 7, 8a-b. 1967 Distacodus stola Lindstrom; Kniipfer, p. 25, pi. 5, figs la-b, 2. 1967 Goniodontus ordovicicus Kniipfer, p. 29, pi. 4, figs 8, 9a-b, 10; pi. 11, fig. 11. 1967 Goniodontus n. sp. Kniipfer, p. 29, pi. 4, fig. 13. 71967 Acodus abnormis Kniipfer, p. 17, pi. 3, figs 9, 13a-b. 1967 Roundya gebersdorfi Kniipfer, p. 35, pi. 4, figs 11-12. 1967 Clavohamulus n. sp. 1 Kniipfer, p. 23, pi. 1, figs la-b, 2a-b. 1967 Clavohamulus n. sp. 2 Kniipfer, p. 23, pi. 1, figs 3a-c. 1967 Lonchodus sp. Kniipfer, p. 34, pi. 8, figs 8-9. 1982 Sagittodontina robusta Kniipfer; Paris et al., p. 21, pi. 2, fig. 1 1 ; pi. 3, figs 1-3, 5; pi. 4, figs 1-2, 4, 6. 1983 Sagittodontina bifurcata Kniipfer; Bergstrom, fig. 4. 1990 Sagittodontina robusta Kniipfer; Fuchs, p. 206, pi. 5, figs 1-8; pi. 7, fig. 1. 71990 Rhode sognathus elegans (Rhodes); Fuchs, p. 206, pi. 6, fig. 4. 1992 Sagittodontina robusta Kniipfer; Bergstrom and Massa, p. 1338, pi. 1, figs 6-8, 79, 10-14, 17. Description. Pa elements of morphogenus ‘ Sagittodontina Kniipfer, 1967’ with cusp slightly inclined posteriorly. The anterior process carries up to five rounded denticles which are suberect or with same inclination as cusp. Denticles are independent of each other. Bifurcation of denticles on posterior process may form immediately posterior to cusp or another denticle, commonly with an opposite inclination to cusp. Two gentle costae, along which denticles of posterior process develop, converge and extend to cusp (PI. 4, fig. 23). No complete specimens have been found, so nature of posterior part is uncertain, but two rows of denticles on the process could have developed more or less parallel and not necessarily connected to each other. Various fragments classified by Kniipfer (1967) as ‘ Lonchodus sp.’ may have occupied this element position (PI. 4, fig. 22). No oral edge has been observed on the platform. The lateral process can be long and weakly denticulated, carrying a maximum of two denticles. A very deep basal cavity is common to all denticles. The denticulation of the anterior process of ‘ Sagittodontus dentatus Kniipfer, 1967’ is more a crenulation of its anterior margin. The platform-like posterior process shows only the posterior part of a narrow and flat area with no visible denticles (PI. 4, fig. 1) or one small denticle (PI. 4, fig. 4). As in the previous morphogenus, two lines converge and extend to the cusp. Pb elements are well preserved and are represented only by sinistral forms. In a few M elements, the cusp appears to split into two small denticles. The posterior process is denticulated and partially preserved only in one specimen (PI. 4, fig. 12). S elements (PI. 4, figs 13-20) are clearly distinguishable, even as fragments, by their cross section; triangular in the Sa and Sb position, but with different symmetry, and square in the Sd. All the elements of our fauna have a very pale colour, on account of their limited wall thickness, and are extremely fragile. Remarks. Sagittodontina robusta is one of the dominant components of our fauna, with well preserved specimens. Thuringia appears to be the best source of specimens of this species. Only elements included by Kniipfer (1967) in the morphogenus ‘ Sagittodontina ’ f S. bifurcata ’, ‘ S. uni- dentata', ' S. robusta ’, ' S. separata') and in the morphospecies ‘ Sagittodontus dentatus Ethington, 1959' develop a platform-like posterior process distally from the cusp, although it is never completely preserved. ‘ Sagittodontus robustus robustus Rhodes, 1953’ and ‘ Sagittodontus robustus flammeus Kniipfer, 1967’ retain their pyramidal shape with no denticles or platform-like expansions (PI. 4, fig. 9). As they are not complete, it is possible that they could have developed a platform similar to the one described for ‘ S. dentatus Ethington, 1959.’ Surprisingly, only sinistral ‘ambalodiform’ elements occur (PI. 4, figs 5-7), as independently noted by Stouge (pers. comm. 1993). In contrast, only dextral forms of another ‘ambalodiform element’, previously considered to belong to Rhodesognathus elegans, are present (PI. 2, figs 1 1-13). It is possible that they belong in the same apparatus, even if not mirror images, and not necessarily in Sagittodontina robusta. Fragments classified as ‘ Clavohamulus n. sp. 1 and 2’ by Kniipfer (1967) are probably part of ramiform lateral processes, as they are small and always broken (PI. 4, fig. 21). Occurrence. Upper Ordovician of Thuringia, Spain, north-west France, ?Bohemia and Libya. FERRETTI AND BARNES: ORDOVICIAN CONODONTS 33 Family icriodellidae Sweet, 1988 Genus icriodella Rhodes, 1953 Type species. Icriodella superba Rhodes, 1953. Icriodella sp. Plate 1, figure 16 Description. Only one fragment of a Pa anterior process has been found in our fauna. The node-like denticles are slightly asymmetrical posteriorly in respect to the axis of the process. A median longitudinal ridge with a ‘zigzag’ pattern runs along the median part of the process. Sides are narrow and the basal cavity deep. Remarks. This distinctive fragment may represent a new species of Icriodella , but owing to its scarcity in our fauna we have preferred to leave it in open nomenclature. Order protopanderodontida Sweet, 1988 Family acanthodontidae Lindstrom, 1970 Genus cornuodus Fahraeus, 1966 Type species. Cornuodus erectus Fahraeus, 1966. Cornuodus longibasis (Lindstrom 1955) Plate 2, figures 18-20 1966 Cornuodus erectus Fahraeus, p. 20, pi. 2, fig. 8a-b; text-fig. 2b. 1974 ‘ Cornuodus ’ longibasis (Lindstrom); Serpagli, p. 43, pi. 7, fig. 2a-b; pi. 20, fig. 12. 1978 Cornuodus longibasis (Lindstrom); Lofgren, p. 49, pi. 4, figs 36, 38^12; text-fig. 25a-c. 1990 Cornuodus longibasis Lindstrom; Stouge and Bagnoli, p. 14, pi. 3, figs 3-7. 1994 Cornuodus longibasis Lindstrom; Dzik, p. 61, pi. 11, figs 8-13; text-fig. 4a. Description. Only six specimens of this form species are present; they are not well preserved. They are bilaterally symmetrical and all with straight profile of the base. Only one (PI. 2, fig. 18) shows longitudinal costae close to the posterior margin. Occurrence. From the lower to upper Ordovician in Europe (Thuringia, Sweden, Poland, Russia), Korea, North and South America. Family drepanoistodontidae Fahraeus and Nowlan, 1978 Genus drepanoistodus Lindstrom, 1971 Type species. Oistodus forceps Lindstrom, 1955. Drepanoistodusl sp. Plate 5, figures 1-12 71959 Drepanodus sp. Lindstrom, p. 439, pi. 3, figs 1-5; text-fig. 3 (3). 1967 Drepanodus n. sp. aff. suberectus (Branson and Mehl); Kniipfer, p. 28, pi. 2, figs 7-8. Description. Erect (PI. 5, figs 3-8) and suberect (PI. 5, figs 1-2, 9-10) elements with large basal cross section. Deep basal cavity. M? elements (PI. 5, figs 11-12) rare and with wide base. Remarks. Drepcmoistodusl sp. is a common species in the fauna. Occurrence. Upper Ordovician of Germany and ?Wales. 34 PALAEONTOLOGY, VOLUME 40 Family protopanderodontidae Lindstrom, 1970 Genus scabbardella Orchard, 1980 Type species. Drepanodus altipes Henningsmoen, 1948. Scabbardella altipes (Henningsmoen, 1948) Plate 1, figures 17-22 1948 Drepanodus altipes Henningsmoen, p. 420, pi. 25, fig. 14. 1967 Acodus flagellus flagellus Knupfer, p. 17, pi. 3, figs 5-8. 1967 Acodus flagellus compactus Knupfer, p. 18, pi. 3, figs 3—4. 1967 Acontiodus altipes Knupfer, p. 19, pi. 3, fig. 1. 1967 Drepanodus flagellus flagellus Knupfer, p. 26, pi. 2, figs 16-18. 1967 Drepanodus flagellus pseudoaltipes Knupfer, p. 26, pi. 2, fig. 11. 1980 Scabbardella altipes (Henningsmoen); Orchard, p. 25, pi. 5, figs 2-5, 7-8, 12, 14, 18, 20, 23-24, 28, 30, 33, 35; text-fig. 4c. 1991 Scabbardella altipes (Henningsmoen); Ferretti and Serpagli, pi. 1, figs 12-14. 1992 Scabbardella altipes (Henningsmoen); Bergstrom and Massa, p. 1339, pi. 1, figs 1, 3^1. 1994 Scabbardella altipes (Henningsmoen); Dzik, p. 64, pi. 11, figs 36-39; text-fig. 6e. Remarks. Scabbardella altipes is a common species in the Thuringian material. Our specimens appear variable in general profile as in length and curvature of the cusp. Both the subspecies (S. altipes subsp. A and B) described by Orchard (1980) are represented. Occurrence. Upper Ordovician of Europe (Great Britain, Spain, France, Carnic Alps, Sardinia, Thuringia, Bohemia, Norway, Sweden), Libya and North America. Order Unknown Family Unknown Genus ISTORINUS Knupfer, 1967 Type species. Istorinus erectus Knupfer, 1967. Istorinus erectus Knupfer, 1967 Plate 5, figures 13-20 1967 Istorinus erectus Knupfer, p. 31, pi. 1, figs 4—6. 1967 Istorinus postdentatus Knupfer, p. 31, pi. 1, fig. 10. 1967 Istorinus recurvus Knupfer, p. 32, pi. 1, figs 7-9. EXPLANATION OF PLATE 4 Figs 1-23. Sagittodontina robusta Knupfer, 1967. 1, IPUM 24790; x 90; upper view of Pa element. 2, IPUM 2479 1 ; x 80 ; lateral view of Pa element. 3, IPUM 24792 ; x 90 ; lateral view of Pa element. 4, IPUM 24793 ; x 70; lateral view of Pa element. 5-7, IPUM 24794; x 75; IPUM 24795; x 85; IPUM 24796; x 95; lateral views of Pb elements. 8-9, IPUM 24797; x 70; IPUM 24798; x 70; lateral views of Pa elements. 10-12, IPUM 24799; x 90; IPUM 24800; x 130; IPUM 24801 ; x 100; lateral views of M elements. 13-14, IPUM 24802; x 75; IPUM 24803; x 60; posterior and lateral views of Sa elements. 15-16, IPUM 24804; x 75; IPUM 24805; x 90; lateral views of Sb elements. 17-18, IPUM 24806; x 100; IPUM 24807; x 90; lateral views of Sc elements. 19-20, IPUM 24808; x 70; IPUM 24809; x 70; lateral views of Sd elements. 21, IPUM 24810; x 100; lateral view. 22, IPUM 24811; x 120; upper view. 23, IPUM 24812; x 145; lateral view of fragment of Pa element showing two thin costae converging to the cusp. PLATE 4 FERRETT1 and BARNES, Sagittodontina robusta 36 PALAEONTOLOGY, VOLUME 40 1967 Drepanodus disymmetricus Kniipfer, p. 26, pi. 2, figs 1-3. 1967 Drepanodus humilis Kniipfer, p. 27, pi. 2, figs 4—6. 1992 Istorinus erectus Kniipfer; Bergstrom and Massa, p. 1338, pi. 1, figs 15-16. Description. Multielement species of Istorinus composed of small, laterally compressed elements with sharp anterior and posterior margins and deep basal cavity expansive in lower two-thirds of cusp and under denticles. Lateral faces are smooth. Denticles are generally present only on one side of cusp, being completely independent from cusp or partially fused with its basal portion. Walls are very thin, hence their pale colour. The cusp is erect to suberect. Three main element morphotypes have been recognized in our fauna. The first type (PI. 5, figs 13-14) has a simple triangular shape with wide base. A small denticle is commonly present on the steeper side of the cusp. In the second type (PI. 5, figs 15, 17-18), which is probably the most familiar form of Istorinus , the basal margin is straight and of variable length. The general profile of the element in lateral view is asymmetrical, having anterior and posterior margins with different curvature, one gentle, the other sharp. One or two denticles, developed parallel or slightly divergent to the cusp axis, are present on the latter edge and may extend to half cusp height. All our specimens bearing two denticles are broken, hence additional denticles may have been present. The gentle margin of the element can be long, being undenticulated or carrying a small denticle at its extremity (PI. 5, fig. 15). A common deep basal cavity extends under the cusp and each denticle. The third type (PI. 5, figs 16, 19) shares the same denticulation, but the basal margin is arched below the cusp. In some specimens, it may be flexed differently in two lateral faces, leaving visible the basal margin of the opposite one. Remarks. Dzik (1989) suggested that Istorinus erectus may simply represent the process fragments of ramiform elements of Sagittodontina robusta. Bergstrom and Massa (1992) noted that their specimens from the upper Ordovician of Libya appear to be complete. Our specimens are also complete (PI. 5, figs 16b, 20), and many are clearly too large to be parts of the ramiform processes. Nevertheless, some posterior processes of the S elements in Sagittodontina robusta are close morphologically, especially to the second morphotype described above. Occurrence. Istorinus erectus is known in the upper Ordovician of Thuringia, Spain, north-west France and Libya. ‘ ambalodiform ’ element Plate 2, figures 1 1-13 1967 Ambalodus triangularis triangularis Branson and Mehl; Kniipfer, p. 20, pi. 9, fig. 2a-b. Description. These ‘ambalodiform’ elements are characterized by a very long, straight and slender denticulated anterior bar. The posterior bar is denticulated and of a different width, being slender or wide. The outer lateral process is not denticulated or carries a single small denticle. Only dextral forms are known. Remarks. We noted above how these forms do not conform to the features of the Pb element of Rhodesognathus elegans. A similar element was included by Savage and Bassett (1986) in EXPLANATION OF PLATE 5 Figs 1-12. Drepanoistodus ? sp. 1, IPUM 24813; x 110; 2, IPUM 24814; x 110; 3, IPUM 24815; x 115; 4, IPUM 24816; x 85; 5, IPUM 24817; x 100; 6, IPUM 24818; x 140; 7, IPUM 24819; x 150; 8, IPUM 24820; x95; 9, IPUM 24821; x 115; 10, IPUM 24822; x 105; 11, IPUM 24823; x 170; 12, IPUM 24824 x 160; all lateral views. Figs 13-20. Istorinus erectus Kniipfer, 1967. 13-14, IPUM 24825; x 105; IPUM 24826; x 120; lateral views. 15, IPUM 24827; x 120; lateral view. 16a-b, IPUM 24828; x 120; x 650; lateral and close oral views. 17, IPUM 24829; x 135; lateral view. 18, IPUM 24830; x 115; lateral view. 19, IPUM 24831; x 120; lateral view. 20, IPUM 24832; x280; oral view. PLATE 5 FERRETT1 and BARNES, Drepanoistodus'!, Istorinus 38 PALAEONTOLOGY, VOLUME 40 Amorphognathus superbus , but as the diagnostic M elements of this species are absent in our material, assignment to A. superbus is unlikely. They have been reported also from Spain (Ferretti 1992). Occurrence. Upper Ordovician of Thuringia and Spain. ‘carniodiform’ element Plate 3. figures 20-22 1955 ‘indeterminate fragment’ Rhodes, pi. 8, fig. 11; pi. 9, fig. 7. 71964 IPrioniodina aflexa Hamar, p. 277, pi. 3, figs 15, 18-19; text-fig. 5. 1967 ‘ Carniodus' sp.; Serpagli, p. 56, pi. 22, figs 1-3. 1967 Carniodus sp. Walliser; Kniipfer, p. 22, pi. 8, figs 6, 10-12. 1980 Eocarniodus gracilis (Rhodes); Orchard, p. 20, pi. 2, figs 19, 30-31, 36. 1982 Carniodus sp. Walliser; Paris et al. , p. 20, pi. 2, fig. 14. Description. Narrow, small elements that develop two processes with up to five irregular denticles that each extend from prominent central cusp. Processes may be completely aligned (PL 3, figs 20, 22) or flexed in an arc (PI. 3, fig. 21). The narrow basal cavity tends to widen slightly under the cusp only in straight specimens. Remarks. The preservation of our material is not good, and the basal ledge that is visible in some elements from Britain (Orchard 1980, pi. 2, fig. 19) has rarely been observed. No complete specimens are present in our collection and we have preferred a morphospecific description. This does not exclude the possibility that they may belong to Eocarniodus gracilis Orchard, 1980. Broken bars of ramiform elements of Amorphognathus have been confused with these elements. This could have occurred also with the ramiform elements of Hamarodus europaeus , which are bigger and closer in size to our specimens. Nevertheless, in no bar of either species did we note the characteristic flexure of the two processes of some elements included here. Occurrence. ?Caradoc and Ashgill of Great Britain, Spain, Bohemia, France, Norway, Sardinia, Carnic Alps and Thuringia; lower Silurian of the Carnic Alps. ‘oistodiform’ element Plate 3, figures 23-24 71967 Oistodus abundans Branson and Mehl; Kniipfer, p. 34, pi. 5, fig. 3. Remarks. Compared with Oistodus niger Serpagli, 1967, the M element of Hamarodus europaeus , some ‘oistodiform’ elements in our fauna are smaller and have a more recurved anterior margin. They also lack the slight concavity of the basal margin at the anterior extremity (the basal margin is therefore convex and only posteriorly concave) and have a more reclined aspect overall. As great variability has been noted above for the corresponding M element of Hamarodus europaeus , it is still possible that they may be an extreme variant within the latter species. Gen. et sp. indet. Plate 2, figure 21 Description. Weakly arched and flexed blade element with subcentral low cusp, a posterior denticulated process and an anterior denticle. Anterior inner process complete and gently expanding in a ridge. Acknowledgements. Our sincere thanks are due to Wolfgang Hammann (University of Wurzburg) for providing us with the Kalkbank sample and relevant stratigraphical details on the German succession, and to FERRETTI AND BARNES: ORDOVICIAN CONODONTS 39 Stig M. Bergstrom (Ohio State University) for helpful discussions. Special thanks go to Enrico Serpagli (Modena University) and Svend Stouge (Geological Survey of Denmark) for valuable comments about various aspects of Ordovician conodont biostratigraphy and taxonomy. Thanks are also extended to anonymous reviewers for expert and accurate revisions. AF is grateful to Lisa Bohach and the SEOS staff for help in all stages of the work and to Diane and Doug Brown for their warm hospitality during her stay in Canada. We are indebted to Claudio Gentilini for his technical assistance with the SEM and Piero Vezzani for his help in the preparation of the plates. CRB acknowledges financial support from the Natural Sciences and Engineering Research Council of Canada. AF’s printing expenses were covered by CNR and MURST grants (and E. Serpagli). REFERENCES aldridge, R. J., purnell, m. a., gabbott, s. E. and theron, J. N. 1995. The apparatus architecture and function of Promissum pulchrum Kovacs-Endrody (Conodonta, Upper Ordovician) and the prioniodontid plan. Philosophical Transactions of the Royal Society of London , Series B, 347, 275-291. — theron, j. n. and gabbott, s. e. 1994. The Soom Shale: a unique Ordovician fossil horizon in South Africa. Geology Today , 10, 218-221. bergstrom, s. m. 1962. Conodonts from the Ludibundus Limestone (Middle Ordovician) of the Tvaren area (S.E. Sweden). Arkiv for Mineralogi och Geologi , 3, 1-61. — 1964. Remarks on some Ordovician conodont faunas from Wales. Acta Universitatis Lundensis , 2, 1-66. 1971. Conodont biostratigraphy of the Middle and Upper Ordovician of Europe and eastern North America. 83-157. In sweet, w. c. and bergstrom, s. m. (eds). Symposium on conodont biostratigraphy. Memoir of the Geological Society of America , 127, 1-499. — 1978. Middle and Upper Ordovician conodont and graptolite biostratigraphy of the Marathon, Texas graptolite zone reference standard. Palaeontology , 21, 723-758. — 1983. Biogeography, evolutionary relationships, and biostratigraphic significance of Ordovician platform conodonts. Fossils and Strata , 15, 35-58. — and massa, d. 1979. Upper Ordovician conodonts from Libya (abs.). IX International Congress Carboniferous Stratigraphy and Geology , Abstract of Papers , 245-246. — 1987. Stratigraphic and biogeographic significance of Upper Ordovician conodonts from Northwestern Libya (abs.). Third Symposium on the geology of Libya, 35. 1992. Stratigraphic and biogeographic significance of Upper Ordovician conodonts from Northwestern Libya. 1323-1342. In Salem, m. j., hammuda, o. s. and eliagoubi, b. a. (eds). The geology of Libya , IV. Elsevier Science Publishers, Amsterdam, 551 pp. — and orchard, m. j. 1985. Conodonts of the Cambrian and Ordovician Systems from the British Isles. 32-67. In higgins, a. c. and Austin, r. l. (eds). A stratigraphical index of conodonts. Ellis Horwood Ltd, Chichester, 263 pp. branson, e. b. and branson, c. c. 1947. Lower Silurian conodonts from Kentucky. Journal of Paleontology , 21, 549-556, pis 81-82. — and mehl, m. G. 1933. Conodonts studies number two. University of Missouri Studies, 8, 77-167. carls, p. 1975. The Ordovician of the Eastern Iberian Chains near Fombuena and Luesma (Prov. Zaragoza, Spain). Neues Jahrbuch fur Geologie and Paldontologie, Abhandlungen, 150, 127-146. cocks, L. R. m. and fortey, r. a. 1988. Lower Palaeozoic facies and faunas around Gondwana. 183-200. In audley-charles, m. g. and hallam, a. (eds). Gondwana and Tethys. Geological Society of London, Special Publication 37. cooper, b. j. 1976. Multielement conodonts from the St. -Clair Limestone (Silurian) of southern Ohio. Journal of Paleontology, 50, 205-217. dzik, j. 1976. Remarks on the evolution of Ordovician conodonts. Acta Palaeontologica Polonica, 21, 395-455. 1989. Conodont evolution in high latitudes of the Ordovician. Courier Forschungsinstitut Senckenberg, 117, 1-28. — 1994. Conodonts of the Mojcza Limestone. In dzik, j., olempska, e. and pisera, a. Ordovician carbonate platform ecosystem of the Holy Cross Mountains. Palaeontologia Polonica , 53, 43-128. epstein, a. g., epstein, j. b. and Harris, l. d. 1977. Conodont color alteration - an index to organic metamorphism. Professional Paper of the United States Geological Survey, 995, I 27. ethington, r. l. 1959. Conodonts of the Ordovician Galena Formation. Journal of Paleontology, 33, 257-292, pis 39-41. 40 PALAEONTOLOGY, VOLUME 40 fahr^eus, l. e. 1966. Lower Viruan (Middle Ordovician) conodonts from the Giillhogen quarry, southern central Sweden. Sveriges Geologiska Undersokning Avhandlingar, 60, 1 — 40. — and hunter, D. R. 1985. Simple-cone conodont taxa from the Cobbs Arm Limestone (Middle Ordovician), New York Island, Newfoundland. Canadian Journal of Earth Sciences , 22, 1171-1182. — and nowlan, g. s. 1978. Franconian (Late Cambrian) to early Champlainian (Middle Ordovician) conodonts from the Cow Head Group, western Newfoundland. Journal of Paleontology, 72, 444-471. ferretti, a. 1992. Biostratigrafia a conodonti del margine settentrionale del Gondwana (Ordoviciano sup.- Ashgill). Unpublished Ph.D. Thesis, University of Modena, 281 pp. — and serpagli, e. 1991. First record of Ordovician conodonts from southwestern Sardinia. Rivista Italiana di Paleontologia e Stratigrafia , 97, 27-34. ELAJS, G. and schonlaub, h. p. 1976. Die biostratigraphische Gliederung des Altpalaozoikums am Polster bei Eisenerz (Nordliche Grauwackenzone, Osterreich). Verhandlungen der Geologischen Bundesanstalt, 2, 257-303. fuchs, a. 1989. Zur Stratigraphie der Grafenthaler Schichten (Ordovizium) in Thiiringen auf der Grundlage von Conodontenuntersuchungen. Verdjfentlichungen Naturhistorisches Museum Schlofi Bertholdsburg Schleusingen , 4, 78-82. 1990. Bemerkungen zu Stratigraphie und Charakter der ordovizischen Conodontenfauna Thiiringens. Neues Jahrbuch fur Geologie und Paldontologie, Monatshefte, 4, 193-214. fuganti, a. and serpagli, E. 1968. Geological remarks on Urbana Limestone and evidence for its Upper Ordovician age by means of conodonts - Eastern Sierra Morena, South Spain. Bollettino della Societa Geologica Italiana , 87, 51 1-521. hafenrichter, m. 1979. Palaontologisch-okologische und Lithofazielle Untersuchungen des " Ashgill-Kalkes” (Jungordovizium) in Spanien. Arbeiten aus dent Paldontologischen Institut Wurzburg , 3, 1-139. hamar, G. 1964. The Middle Ordovician of the Oslo region, Norway. 17. Conodonts from the lower Middle Ordovician of Ringerike. Norsk Geologisk Tidsskrift, 44, 243-292. 1966. The Middle Ordovician of the Olso Region, Norway. 22. Preliminary report on conodonts from the Olso-Asker and Ringerike districts. Norsk Geologisk Tidsskrift, 46, 27-83. hammann, w., robardet, m. and romano, m. 1982. The Ordovician System in southwestern Europe (France, Spain, Portugal). Publication of the International Union of Geological Sciences, 11, 1-47. hartevelt, j. 1970. Geology of the upper Segre and Valira valleys, central Pyrenees. Leidse Geologische Mededelingen, 45, 167-236. hass, w. h. 1959. Conodonts from the Chappel Limestone of Texas. Professional Paper of the United States Geological Survey, 294J, 365—400. havlicek, v. 1976. Evolution of Ordovician brachiopod communities in the Mediterranean Province. 349-358. In BASSETT, M. G. (ed.). The Ordovician System: proceedings of a Palaeontological Association symposium, Birmingham, September 1974. University of Wales Press and National Museum of Wales, Cardiff, 696 pp. 1977. Brachiopods of the Order Orthida in Czechoslovakia. Rozpravy Ustredniho Ustavu Geologickeho, 44, 1-327. helmcke, d. and rock, G. 1974. Zur Altersstellung der Porphyroide in der Lagerstattenprovinz Sarrabus- Gerrei. Zeitschrift der Deutschen Geologischen Gesellschaft, 125, 91-98. henningsmoen, g. 1948. The Tretaspis Series of the Kullatorp Core. Bulletin of the Geological Institutions of the University of Uppsala, 32, 374—432. jaeger, h., havlicek, v. and schonlaub, h. p. 1975. Biostratigraphie der Ordovizium/Silur-Grenze in den Siidalpen-Ein Beitrag zur Diskussion um die Hirnantia- Fauna. Verhandlungen Geologische Bundesanstalt, 4, 271-289. khodalevich, a. n. and tschernich, v. v. 1973. [New subfamily Belodellinae (conodonts).] Trudy Sverdlovskogo Gornogo Instituta, 93, 42 — 47. [In Russian]. knupfer, j. 1967. Zur Fauna und Biostratigraphie des Ordoviziums (Grafenthaler Schichten) in Thiiringen. Freiberger Forschungshefte, C220, 1-119. leone, f., hammann, w., laske, r., serpagli, e. and villas, e. 1991. Lithostratigraphic units and biostratigraphy of the post-sardic Ordovician sequence in south-west Sardinia. Bollettino della Societa Paleontologica Italiana, 30, 201-235. lindstrom, m. 1955. Conodonts from the lowermost Ordovician strata of south-central Sweden. Geologiska Foreningens i Stockholm Forhandlingar, 76, 517-603. 1959. Conodonts from the Crug Limestone (Ordovician, Wales). Micropaleontology, 5, 427-452. 1970. A suprageneric taxonomy of the conodonts. Fethaia, 3, 427-445. FERRETTI AND BARNES: ORDOVICIAN CONODONTS 41 lindstrom, m. 1971. Lower Ordovician conodonts of Europe. 21-61. In sweet, w. c. and Bergstrom, s. m. (eds). Symposium on conodont biostratigraphy. Memoir of the Geological Society of America , 127, 1^199. 1977. Genus Amorphognathus Branson and Mehl 1 9 3 3 f . 21-52. In ziegler, w. (ed. ). Catalogue of conodonts , Vol. 3. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart, 574 pp. — and pelhate, a. 1971. Presence de conodontes dans les calcaires de Rosan (Ordovicien Moyen a Superieur, Massif Armoricain). Memoires du Bureau de Recherches Geologiques et Minieres , 73, 89-91. — racheboeuf, p. R. and henry, j.-l. 1974. Ordovician conodonts from the Postolonnec Formation (Crozon Peninsula, Massif Armoricain) and their stratigraphic significance. Geologica et Palaeontologica , 8, 15-28. lofgren, a. 1978. Arenigian and Llanvirnian conodonts from Jamtland, northern Sweden. Fossils and Strata , 13, 1-129. marek, L. and havlicek, v. 1967. The articulate brachiopods of the Kosov Formation (Upper Ashgillian). Vestnik Ustfedniho Ustavu Geologickeho , 42, 275-284. martin, F. 1982. Some aspects of late Cambrian and early Ordovician acritarchs. 29-40. In bassett, m. g. and dean, w. T. (eds). The Cambrian-Ordovician boundary : sections, fossil distributions , and correlations. National Museum of Wales, Geological Series, 3, Cardiff, 227 pp. nowlan, G. s. 1983. Biostratigraphic, paleogeographic, and tectonic implications of Late Ordovician conodonts from the Grog Brook Group, northwestern New Brunswick. Canadian Journal of Earth Sciences, 20, 651-671. orchard, m. j. 1980. Upper Ordovician conodonts from England and Wales. Geologica et Palaeontologica, 14, 9-44. Paris, F. 1981. Les chitinozoaires dans le Paleozoique du sud-ouest 1’ Europe. Memoir e de la Societe Geologique et Mineralogique de Bretagne, 26, 1—412. — pelhate, A. and weyant, m. 1982. Conodontes ashgilliens dans la Formation de Rosan, coupe de Lostmarc'h (Finistere, Massif armoricain). Consequences paleogeographiques. Bulletin de hi Societe Geologique et Mineralogique de Bretagne , 13, 15-35. Rhodes, F. h. t. 1953. Some British Lower Palaeozoic conodont faunas. Philosophical Transactions of the Royal Society of London, Series B, 237, 261-334. — 1955. The conodont fauna of the Keisley Limestone. Quarterly Journal of the Geological Society, London, 111, 1 17-142, pis 7-10. robardet, m. and dore, f. 1988. The Late Ordovician diamictic formations from southwestern Europe: north- Gondwana glaciomarine deposits. Pa/aeogeography, Palaeoclimatology, Palaeocology, 66 (for 1987), 19-31. — Paris, F. and racheboeuf, p. r. 1990. Palaeogeographic evolution of southwestern Europe during Early Palaeozoic times. 411—419. In McKERROW, w. s. and scotese, c. r. (eds). Palaeozoic pa/aeogeography and biogeography. Geological Society Memoir 12. sansom, i. j., Armstrong, H. A. and smith, M. p. 1995. The apparatus architecture of Panderodus and its implications for coniform conodont classification. Palaeontology, 31, 781-799. sarmiento, g. n. 1990. Conodontos de la Zona ordovicicus (Ashgill) en la Caliza Urbana, Corral de Calatrava (Ciudad Real). Geogaceta, 7, 54-56. savage, n. m. and bassett, m. g. 1986. Caradoc-Ashgill conodont faunas from Wales and the Welsh Borderland. Palaeontology , 28, 679-713. schonlaub, h. p. 1969. Das Palaozoikum zwischen Bischofalm und Hohem Trieb (Zentrale Karnische Alpen). Jahrbuch der Geologischen Bundesanstalt, 112, 265-320. — 1971. Palaeo-environmental studies at the Ordovician/Silurian boundary in the Carnic Alps. Memoires du Bureau de Recherches Geologiques et Minieres, 73, 367-378. — 1979. Das Palaozoikum in Osterreich. Abhandlungen der Geologischen Bundesanstalt , 33, 1-124. — (ed.) 1980. Second European Conodont Symposium. Guidebook, Abstracts. Abhandlungen der Geologischen Bundesanstalt , 35, 1-213. serpagli, e. 1967. 1 conodonti dell’Ordoviciano superiore (Ashgilliano) delle Alpi Carniche. Bollettino della Societd Paleontologica Itqliana, 6, 30-111. — 1974. Lower Ordovician Conodonts from Precordilleran Argentina (Province of San Juan). Bollettino della Societd Paleontologica Italiana, 13, 17-98. — and greco, a. 1965. Osservazioni preliminari su alcuni Conodonti ordoviciani e siluriani delle Alpi Carniche italiane. Bollettino della Societd Paleontologica Italiana, 3 (for 1964), 192-211. stauffer, c. R. 1935. Conodonts of the Glenwood Beds. Bulletin of the Geological Society of America , 46, 125-168. 42 PALAEONTOLOGY, VOLUME 40 steiner, J. and falk, f. 1981. The Ordovician Lederschiefer of Thuringia. 579-581. In hambrey, m. j. and harland, w. b. (eds). Earth's pre-Pleistocene glacial record. Cambridge University Press, Cambridge, 1004 pp. storch, p. 1990. Upper Ordovician-lower Silurian sequences of the Bohemian Massif, central Europe. Geological Magazine , 127, 225-239. stouge, s. and bagnoli, g. 1990. Lower Ordovician (Volkhovian-Kundan) conodonts from Hagudden, northern Oland, Sweden. Palaeontographia Italica , 77, 1-54. sweet, w. c. 1988. The Conodonta : morphology, taxonomy, paleoecology and evolutionary history of a long- extinct animal phylum. Oxford University Press, New York, 212 pp. and bergstrom, s. m. 1976. Conodont biostratigraphy of the Middle and Upper Ordovician of the United States Midcontinent. 121-151. In bassett, m. g. (ed. ). The Ordovician System, proceedings of a Palaeontological Association symposium, Birmingham , September 1974. University of Wales Press and National Museum of Wales, Cardiff, 696 pp. 1984. Conodont provinces and biofacies of the Late Ordovician. Special Papers of the Geological Society of America, 196, 69-87. theron, j. n., rickards, R. b. and aldridge, r. j. 1990. Bedding plane assemblages of Promissum pulchrum , a new giant Ashgill conodont from the Table Mountain Group, South Africa. Palaeontology, 33, 577-594. vavrdova, M. 1974. Geographical differentiation of Ordovician acritarch assemblages in Europe. Review of Palaeobotany and Palynology, 18, 171-175. viira, v. 1974. [ Ordovician conodonts of the Baltic Region .] ‘Valgus’, Tallinn, 142 pp. [In Russian], walliser, o. h. 1964. Conodonten des Silurs. Abhandlungen der Hessischen Landesamtes fur Bodenforschung zu Wiesbaden, 41, 1-106. weyant, m., dore, f., le gall, j. and poncet, j. 1977. Un episode calcaire ashgillien dans Test du Massif armoricain; incidence sur Page des depots glacio-marins fini-ordoviciens. Comptes Rendus de I'Academie des Sciences de Paris, 284, 1147-1149. Whittington, h. b. and hughes, c. p. 1972. Ordovician geography and faunal provinces deduced from trilobite distribution. Philosophical Transactions of the Royal Society of London, Series B, 263, 235-278. young, t. p. 1989. Eustatically controlled ooidal ironstone deposition: facies relationships of the Ordovician open-shelf ironstones of Western Europe. 51-63. In young, t. p. and taylor, w. e. g. (eds). Phanerozoic ironstones. Geological Society Special Publication, 46. 1992. Ooidal ironstones from Ordovician Gondwana: a review. Palaeogeography , Palaeoclimatology, Palaeoecology , 99, 321-347. ANNALISA FERRETTI Department of Earth Sciences Via Universita 4 41100 Modena, Italy CHRISTOPHER R. BARNES School for Earth and Ocean Sciences University of Victoria P.O. Box 3055 Victoria B.C. V8W 3P6 Canada Typescript received 9 January 1995 Revised typescript received 6 October 1995 MID MESOZOIC FLORAS AND CLIMATES by R. N. L. B. HUBBARD and U. C. BOULTER Abstract. Multivariate statistical analyses of pollen and spore assemblages from Late Triassic to Early Cretaceous sedimentary strata in north-west Europe and earlier Jurassic deposits in Western Australia have allowed the identification of three major climatic-ecological groupings in each area. In each hemisphere the sporomorph groupings are interpreted as reflecting semi-tropical, cool-temperate, and intermediate climatic conditions. Only about 8 per cent, of the pollen and spore taxa occur in both hemispheres; among these, there is a high degree of consistency in climatic-ecological behaviour between north and south. When the constituents of the groups are plotted as summary pollen diagrams in the conventional Quaternary palynological manner, climatic ‘fingerprints' emerge which allow correlations of hitherto unprecedented accuracy, whether within a region or between hemispheres. The changing proportions of the different groupings allow the first pollen-based quantitative reconstructions of Jurassic climates to be made. Uniformitarianism is one of the foundations of geology. We present the results of a decade- long study of Jurassic palynology, in which techniques that were applied successfully to elucidate Cenozoic palynology (Hubbard and Boulter 1983) are applied to Lower and Middle Jurassic pollen (strictly, miospore) analyses. In both cases, multivariate statistical methods were used to find the natural patterns of association in the pollen analyses, and the climatic conditions associated with the various vegetations were deduced. Summary diagrams displaying the interplay of the mjaor vegetational-climatic groupings then allow correlation with the global temperature record which the diagrams reflect. In the Tertiary, pollen diagrams can be correlated with oxygen isotope records and dinoflagellate cyst diagrams (Hubbard et al. 1994); in the Jurassic, the oxygen isotope record cannot be used, but correlations between Australia and Europe, and with dinoflagellate cyst diagrams, leave little doubt that global temperature proxies are likewise involved. Quantitative palynology, in the form of pollen diagrams, is arguably the single most powerful tool of Quaternary palaeoecologists, and is regularly used for correlation and vegetational reconstruction at local to global scales. That palynology is used in analogous ways less and less in earlier geological periods reflects the fact that ecological understanding is almost essential for such treatment. We report here the results of multivariate statistical analyses of Late Triassic to Early Cretaceous pollen assemblages from Australia and Europe. From these analyses we have produced objective syntheses of ecological and climatic developments, and present the first continuous, quantitative reconstructions of Late Triassic and earlier Jurassic palaeoecology and climate. Our pollen diagrams give a global perspective over about 40 million years, for a period the climatic history of which has hitherto been largely speculative. SOURCES OF DATA Our study has twin foundations at what are now opposite ends of the globe, but which were considerably closer between 100 and 200 million years ago. Filatoff (1975) has published 246 detailed, high quality palynological analyses from some deep boreholes in Western Australia, which in earlier Jurassic times was at a latitude of about 45° S on the eastern seaboard of the temperate zone of Pangaea (Text-fig. 1 ). North-west Europe occupied an east-central sub-tropical position (30° N) in the super-continent, about 70° west of Western Australia. In later Jurassic-Early Cretaceous times the development of the precursor of the Atlantic Ocean took place to the south- west, forming a large sub-tropical sea. [Palaeontology, Vol. 40, Part 1, 1997, pp. 43-70] © The Palaeontological Association 44 PALAEONTOLOGY, VOLUME 40 1 Gun Island 7 Karindal 13 Yon’s Nab 2 Hill River 8 Kendelbach 14 Cambridge 3 Badaminna 9 Hasty Bank 15 Upton 4 Cockbum 10 Cloughton Wyke 16 Lavemock 5 Valhall 11 Gristhorpe 17 Lame 6 Vilhelmsfalt 12 Scalby Ness 18 Scoresby text-fig. 1. Palaeogeographical map showing the Earth in early Jurassic times (Europe and east Greenland enlarged in the inset on the right), and the locations of the critical sections and boreholes discussed in this analysis. Upland areas are stippled, ancient coastlines are shown with a solid line, and modern coastlines (or plate margins) are pecked. Pollen analytical investigations of Mesozoic strata have been carried out in Europe over four decades by a number of microscopists (Couper 1958; Norris 1963; Muir 1964; Guy 1971; Wilkinson 1978; Guy-Ohlson 1981, 1982, 1990; Riding 1983). The assembling of a 712-taxon master list of pollen and spore types was an essential preliminary stage in the analysis of the European data. From this list, 153 generally recognizable form-genera were isolated (Boulter and Windle 1993). As the analysis proceeded, this definitive classification was reduced by further lumping and synonymy to an irreducible core of 73 Jurassic pollen and spore taxa, upon which our conclusions are based. An additional corpus of related data involved Late Triassic pollen analyses from Europe and Greenland (Orbell 1972, 1973; Morbey 1975; Lund 1977), which at that time were at about 40° N on the north-west corner of Pangaea. These additional Triassic data added a further 36 palynomorph taxa to the master list. ANALYTICAL METHODS The analytical methods employed in this study are essentially the same as used earlier to elucidate Tertiary plant ecology (Hubbard and Boulter 1983). The pollen counts were turned into percentages, and these were used to construct a matrix of Pearson’s product-moment correlation coefficients, which (in turn) was investigated by principal components and cluster analyses. They are standard techniques of Quaternary palaeoecology (see, for instance, Birks and Birks 1980), and are much the same as those employed by Imbrie and Kipp (1971) in studying the climatic significance of Pleistocene foraminifera, a work that inspired our studies. HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 45 Our primary analytical tool was principal components analysis, which defined the associations implied in the correlation matrix. Examination of the co-ordinate plots of the loadings of the (un- rotated) principal components, the stratigraphical distributions of the scores of the rotated principal components, and the cluster analyses allowed us to identify three major groups of pollen and spore types in each hemisphere: S I, S II and S III in the southern hemisphere, N I, N II and N III in the northern. Rotated principal components for which samples from different localities score heavily in the same chronological period tend to be related climatically. Taxa sharing climatic-ecological affinities tend to be grouped together in the co-ordinate plots of the unrotated principal components, and in the cluster analyses. By checking the implications of each of these lines of evidence against each other, misattributions can be identified and eliminated. Tables 1 and 2 give the composition of the first four unrotated principal components from the Australian and European data-sets respectively; the figures describe the overall structures inherent in the data. The groupings are indicated in the dendrograms (Text-figs 2-3) which illustrate these relationships; the groupings are derived from the principal components analyses, and the dendrograms are merely a convenient and readily-comprehended way of summarizing the analytical results. It should be remembered that the dendrograms are two-dimensional representations of the complex patterns inherent in the correlation matrices. The simplification involved is one reason why the patterns in the dendrograms often deviate from the mathematically controlled truths of the principal components solutions. The composition of the Australian and European rotated principal components is outlined in Tables 3 and 4. In the northern hemisphere, the most conspicuous patterns involve groups of tropical spores and pollen that were particularly well represented in later Jurassic and Early Cretaceous times (Group N III, Text-fig. 3); and an opposing group that recorded a dramatic, intense, and extended cold episode around the Triassic-Jurassic (Rhaetian-Hettangian) boundary (Group N I, Text-fig. 3). The southern intermediate group (S II, Text-fig. 2) is particularly well represented in the Australian earlier Jurassic strata, whose conspicuously red colour suggested to Filatoff (1975) that they were deposited in either a tropical or an arid environment. As the representation of the tropical Group S III is trivial, the arid alternative is indicated. This is of considerable relevance to the northern hemisphere results, where the analogous group is common - in some cases (not considered here), almost certainly misleadingly over-represented - in the Late Triassic and earlier Jurassic sequences, but becoming much rarer in the later Jurassic and Early Cretaceous. The palynological consistency, and steady sediment accumulation rates (attested by the age- verms-depth plots) in many of the Australian borehole analyses allow the best sections of their records (Badaminna 1, Gun Island 1, and Cockburn 1) to be overlapped, to give a long and detailed climatic history of the earlier Jurassic (Text-fig. 4), once allowance had been made for identified unconformities and changes in sedimentation rates. Text-figure 5 shows the age-vcrx«x-depth plots for the three crucial sections: the inconsistencies in the most closely controlled parts are about 100000 years. Text-figure 4 was compiled using the results from Cockburn 1 from 155 to 166 Ma, Badaminna 1 to 177 Ma, and then Gun Island 1. In the oldest parts of the sequences (earlier than about 184 Ma) the stratigraphical control degrades rapidly. The upper part of the Badaminna 1 sequence (above 650 m) seemed anomalous: the oscillations in the pollen diagram were inconsistent with the records from FilatofTs other sequences covering this period. It was therefore rejected. To preserve integrity, no points were interpolated from another core where the basic shape of the pollen curves would have changed in consequence. The top of the Australian sequences were believed to be approximately Callovian-Oxfordian, while certain central parts were associated with ammonite faunas that linked them with the earliest part of the European Bajocian (Filatoff 1975). In the palynological dendrograms (Text-figs 2-3) we have attempted to relate pollen and spore types to extinct groups of plants in the hopes of clarifying the ‘story’ the diagrams tell. This exercise is highly contentious for two reasons: (1) the higher levels of classification in all taxonomic schemes are increasingly abstract intellectual speculations, involving increasing proportions of interpret- ation; and (2) the relationship between taxonomic structures and pollen and spore morphology is notoriously complicated, and each palynologist dealing with extinct plants will tend to have their 46 PALAEONTOLOGY, VOLUME 40 table 1 . The first four unrotated principal components yielded by the Australian data. They account for 19-5 per cent, of the variance in the correlation matrix. Principal component: 12 3 4 Stereisporites psilatus 0-32 -013 001 -0-08 Stereisporites cintiquasporites 0 55 -003 006 -016 Rogalskaisporites cicatricosus 0-48 -0-24 001 -0-22 Rogalskaisporites canaliculus 0-76 -016 001 -018 Polycingulatisporites crenulatus 0-47 -0-23 -001 -0-29 Polycingulatisporites striatus 010 -008 -Oil -016 Antulsporites varigranulatus 0-39 -0-40 -0-14 -013 Antulsporites clavus 0-20 009 0 -009 Antulsporites saevus 0-29 -0-36 0-51 0 41 Foveosporites moretonensis 0-54 -016 -012 -0-14 Staplinsporites caminus 013 -002 -014 -0-17 Staplmsporites telatus 0-44 0-50 0-14 -001 Staplinsporites mathurii 0-22 -004 -006 -0-20 Staplinsporites perforatus 0-47 018 0-08 -0-04 Foveosporites multifoveolatus 0-37 013 005 -006 Foveotriletes cf. irregulatus -003 -012 -014 -009 Densoisporites circumundulatus 009 -012 -003 -010 Densoisporites sp. A -005 0 -0-04 -005 Densoisporites sp. B 016 — 0-31 -013 001 Lycopodiacidi tes cern iidi tes 0-20 -0-21 -0-10 -Oil Camarozonosporites ramosus 0-63 -004 Oil -0-26 Camarozonosporites clivosus 0-72 0-25 0-17 -0-25 Leptolepidites major 0-30 -0-25 -013 -006 Uvaesporites equibossus 016 -0-29 -013 -006 Uvaesporites crassibalteus 0-24 -0-27 -009 003 Uvaesporites sp. -001 002 -001 0-04 Lophotriletes sp. 0-25 -0-34 -0-17 -0-07 Apiculatosporites 0-43 0-20 002 -0-07 Acanthotriletes levindensis 0-22 0-34 -0-38 0-56 Conbaculatisporites cf. mesozoicus 004 — 0-33 0-56 0-46 Neoraistrickia taylorii 012 -009 -012 -0-08 Neoraistrickia pilobaculata 0-62 006 009 -0-08 Neoraistrickia trichosa 0-51 0-28 -0-34 0-40 Neoraistrickia densata 0-63 014 0-17 -0-22 Neoraistrickia truncata 0-61 014 0-05 -0-08 Neoraistrickia sp. A 0-32 -0-37 0-56 0-33 Neoraistrickia spp. 0-54 012 -0-28 0-36 Lycopodiumsporites rosewoodensis -0-17 0 -0-11 -008 Lycopodiumsporites reticulumsporis 0-34 -013 -013 -015 Lycopodiumsporites austroclavatis 0-39 010 -015 0-37 Lycopodiumsporites circolumensis 0-64 019 Oil -013 Lycopodiumsporites sp. A -012 010 001 -016 Lycopodiumsporites spp. 0-58 -0-04 010 -018 Dictyosporites speciosus 0-30 -002 009 -0-07 Dictyosporites complex 0-41 0-30 -0-37 0-43 Paxillitriletes -Oil 010 0 -003 Pilasporites marcidus -016 -004 -0-03 -002 Calamospora mesozoica -012 003 -002 -009 Marattisporites scabratus -012 -0-07 005 006 Todisporites minor -016 0-05 015 -003 Osmundacidites wellmanii 0-20 -008 0-47 0-26 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 47 TABLE 1. ( COUt .) Principal component: 12 3 4 Baculatisporites comaumensis 013 -0-22 -0-09 -007 Verrucosisporites varians 010 — 0 21 — 0T7 -008 Verrucosisporites rugulodomus -015 004 -006 -007 Rugulatisporites chamiomatus -018 010 0-07 -016 Cyathidites australis 012 004 007 -004 Cvathidites minor -005 -015 002 009 Cibotiumspora juriensis -0-06 -013 -009 -0-20 Dictyophy llidites equiexinus 0-23 002 007 002 Dictyophyllidites harrisii -0-23 016 Oil -012 Matonisporites crassiangulatus -012 010 003 -015 Anapiculatisporites clausonensis -005 0-27 0-07 -018 Foramifiisporites tribulosus -013 010 004 -013 Gleicheniidites senonicus 0-48 003 015 013 Duplexsporites problematicus -0-03 008 -0-53 0-40 Contignisporites cooksonii 0-58 0-54 0T4 -003 Trilobosporites antiquus 012 -012 -0-10 005 Ischyosporites cratens 005 009 -009 -002 Ischvosporites vollcheimeni 009 -009 -003 001 Ischyosporites marburgensis -012 001 0-04 006 Klukisporites variegatus -0-29 Oil 010 -005 Klukisporites neovariegatus 005 0 001 002 Klukisporites lacunus 002 016 -Oil 0-04 Klukisporites scaberis 0-44 -Oil 0 -006 Lygodioisporites perverrucatus 016 -0-28 -008 005 Retusotriletes mesozoicus -003 -01 1 -Oil 0 Granulatisporites sp. 002 -016 -013 -001 Converrucosisporites variverrucatus 0-20 -017 -010 -0-04 Cadargasporites baculatus -008 0-34 0-29 001 Cadargasporites reticulatus -010 016 0-27 -003 Nevesisporites vallatus -015 014 Oil -013 Murospora florida 0-50 015 014 -010 Lycopodiacidites asperatus 0-46 0-32 009 001 Monolites couperi -006 0 -004 -003 Reticuloidosporites sp. 0-47 — 0-33 -005 -0-27 Obtusisporites yarragadensis 0-26 0 41 006 009 Obtusisporites cf. canadensis 018 018 -001 007 Cycadopites follicularis Oil -0-27 0-54 0-32 Vitreisporites pallidus -0-07 -005 007 005 Su/cosaccispora lata -0-22 018 006 -0-25 Alisporites lowoodensis -002 -008 0-23 0-23 Alisporites cf. grandis -007 003 Oil -003 Pinuspollenites parvisaccatus 0-34 -004 -018 0-27 Pinuspollenites globosaccus 0-24 -003 -0-24 0-43 Podocarpidites ellipticus 004 0-29 004 0-27 Podocarpidites cf. verrucosus Oil -0-22 -018 012 Podocarpidites cf. multesimus 0-27 -013 007 -003 Podosporites castellanosii 018 017 -003 0 31 Araucariacites australis 007 -008 -0-34 -013 Callialasporites segment atus -016 0-20 019 -005 Callialasporites minus -007 -005 0-48 019 Callialaspirites microvelatus -012 -004 -Oil -017 Callialasporites dampieri 009 -0-07 -0-32 006 Callialasporites turbatus -0-27 -0-04 -0-32 005 48 PALAEONTOLOGY, VOLUME 40 c 'c '5 c ! uuu 6: i « ^ i i. ^ D.a.g-a.0. a. c- ( i-g tit II t|s ?c/3 CQ J CD _T_T CQ JjV 1 j? J.-| I 2 g. g.-g -g ■§ -D-"| ■g.-f 2 j a. Q. Q.-g g •§ d. a. o.-§ Q.-§ a. | — — — — — — c n C/3 C/3 C/3 text-fig. 2. For caption see opposite. table 1. ( coni .) Principal component : 1 2 3 4 Callialasporites trilobatus 0-35 027 -0-03 -003 Classopollis chateaunovi -046 013 0-08 -020 Classopollis simplex -018 0-02 006 -010 Classopollis cf. meyeriana -006 -002 -013 -013 Classopollis anasillos -0-08 0-06 010 -004 Exesipollenites tumulus -027 019 015 -017 Exesipollenites scabratus -Oil 006 -000 -008 Proportion of total variance (per cent.) 9-29 3-54 3-46 3-18 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 49 — — — — — C/0 C/0 C/0 C/0 CO C/0 C/0 C/0 CO text-fig. 2. Average Link dendrogram showing the relationships between the 112 Australian Jurassic pollen and spores in the 246 samples involved in this study. Sporomorphs of essentially cold environments (identified from the principal components analyses) are designated as Group S I, thermophilous ones are classified as Group S III; and Group S II contains the climatically intermediate taxa, which are also apparently associated with arid conditions. The family attributions are those of Filatoff (1975). table 2. The first four unrotated principal components yielded by the European data. They account for 15-7 per cent, of the variance in the correlation matrix. Principal component: 12 3 4 Stereisporites stereoides 0-36 -0-28 -0-08 -009 Stereisporites antiquasporites 0-23 -014 0-15 0-28 Sculptisporis seeburgensis -004 -001 0-04 003 Foveasporis irregularis -012 -044 0-55 -0-20 Polycingulatisporites cir cuius 0-45 -014 006 -0-17 Staplinisporites caminus -0-15 -0-37 -0-25 -0-09 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 50 PALAEONTOLOGY, VOLUME 40 TABLE 2. ( cont .) Principal component : 12 3 4 Densoisporites velatus -010 -0-02 0-27 0 18 Neoraistrickia gristhorpensis -015 — 0-16 0 31 0-21 Lycopodiacidites rugulatus -0-08 -0-21 0-26 0 Lycopodiacidites cerniidites 0-38 -0-23 002 — 0 13 Lycopodiacidites rhaeticus 007 -0-03 0-04 0-07 Retitriletes pseudoreticulatus 012 — 0-21 0-29 0-47 Retitriletes austroclavatidites -006 -0-34 — 0-17 — 0-10 Retitriletes clavatoides -009 — 018 0 12 0-50 Foveotriletes microreticulatus -0-07 — 0-11 0-17 0-14 Lycopodiumsporites sp. A -005 — 0 15 0-20 0-48 Rotverrusporites major -009 — 0 15 -0-07 — 0-13 Rotverrusporites equatibossus -014 -0-46 0-59 — 0 16 Rotverrusporites bossus -012 — 0 16 0-29 -0-05 Leptolepidites plurituberosus -010 -0-43 -006 -0-25 Contignisporites glebulentus 0-07 -0-32 0-07 -005 Aequitriradites spinulosus -001 — 0-10 -0-07 -0-06 Vallizonosporites pseudoalveolatus -010 — 0-17 -003 — 0 01 Cicatricosisporites brevillaesurae -008 — 0-12 -0-22 0-02 Couperisporites complexus -008 -002 -0-05 0-02 Duplexisporites problematicus -010 -0-32 0-06 -0-05 Heliosporites reissingeri 0-35 — 0 10 002 -0-17 Calamospora mesozoica 0-24 -0-05 0-07 0 01 Marattisporites scabratus 0-02 — 0-11 002 0-32 Todisporites major + minor -0-64 0-12 0 0-27 Osmundacidites wellmanii 0-23 -0-23 009 0-16 Pilosisporites brevipappillosus -005 — 0 10 — 0 14 0 Concavisporites crassexinus 0-42 — 0-12 0 13 0-12 Trilobosporites bernissartensis -009 -0-40 -0-29 — 0 19 Concavissimisporites variverrucatus 0-28 -0-38 -008 — 0-18 Deltoidospora australis -0-22 -002 016 0-24 Deltoidospora concavus -0-04 0 0-25 0-02 Deltoidospora neddeni 0-72 — 018 0 10 -0-04 Deltoidospora equiexinus -013 -0-27 -0-22 0 15 Dictyophyllidites harrisii -016 0 0 13 0-03 Camarozonosporites rudis 0-46 -0-15 006 — 0 10 Krauselisporites linearis -0-10 006 0 14 -0-07 Endosporites rhysoseus -0-06 -0-25 — 0 17 — 0-17 Endosporites jurassicus -Oil -0-38 0-49 — 0 14 Matonisporites crassiangulatus -0-05 -002 0-01 0 13 Cibotiumspora sinuata — 0-21 -0-06 0-41 -004 Cibotiumspora tricuspidata -0-04 002 003 -0-03 Obtusisporis canadensis — 014 0-08 0 21 — 015 Gleicheniidites senonicus — 0-13 -0-20 — 0-31 0 Ischyosporites lacunus -0-24 -0-40 0-34 -0-23 Trachysporites asper 001 0 003 010 Undulatisporites concavus — 0 14 — 0-33 0-48 006 Triancoraesporites ancorae 0-48 0 0-09 -0-07 Platyptera trilingua -004 -0-05 008 004 Convolutispora microrugulatus 0-49 — 0 12 007 — 0 14 Auritulinasporites scanicus -0-04 — 0-01 0 12 0-03 Anapiculatisporites cronaferreus 0-26 -0-28 -004 — 0 16 Uvaesporites argentaeformis 0 01 0 12 0-22 -007 Nevesisporites bigranulatus — 0-13 0 1 1 0 12 — 0-12 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 51 TABLE 2. (COIlt.) Principal component: Laevigatosporites couperi -016 -0-28 0-49 -0-21 Lygodioisporites perverrucatus -0-07 0-04 003 -002 Plicatella tricornitata -0-02 -0-27 -016 -017 Reticulisporites venulosus -0-05 -013 -Oil -0-05 Cicatricosisporites dorogensis -0-15 -0-32 -042 -0-06 Murospora spp. -0-03 -0-25 -Oil -012 Clavifera triplex 0-17 -0-05 004 0-21 Zebrasporites spp. 0-52 -008 009 -01 I Classopollis torosus -Oil 0-24 0-20 -005 Classopollis reclusus -012 -001 -0-35 -003 Classopollis meyeriana 0-47 0-21 0-04 -010 Inaperturopollenites australis -0-37 012 0-29 -010 Chasmatosporites spp. 012 0-02 0-08 016 Schizosporis spriggi -006 -0-25 -018 -010 Peltandripites tener -003 -004 -0-09 -001 Undulatasporites callosus -001 0-01 -0-03 0 Naiaditaspora anglica 0-27 -Oil 002 -013 Cycadopites carpentieri -010 -0-03 0-08 0-08 Cycadopites subgranulosus 015 -005 004 0-21 Cycadopites minimus 0-15 -0-07 0-07 0-38 Clavatipollenites hughesii -008 -0-09 -0-22 0-03 Eucommiidites troedssonii -0-06 -0-20 -015 0-02 Phyllocladites microreticulatus -005 -0-05 -003 -01 1 Walchites sp. -012 Oil 014 -014 Vitreisporites pallidus 008 -019 -013 0-17 Spheripollenites scabratus -0-08 -004 -006 0 21 Spheripollenites psilatus -0-24 0-22 0-21 -0-23 Pinuspollenites minimus -0-08 -0-18 -019 -0-09 Alisporites dunrobinensis 0-26 001 Oil 0-29 Alisporites microsaccus Oil 0-46 0-24 — 0-21 Alisporites thomassii 0-45 0-19 0-21 017 Alisporites robustus -003 -013 016 0-40 Pityosporites similis -0-30 -006 -0-35 -010 Pityosporites microalatus -0-23 -001 -019 004 Pityosporites scaurus 004 -Oil 012 0-31 Polonisaccus ferrugineus -002 0-03 009 003 Parvisaccites enigmatus 0-40 -012 012 -0-07 Platysaccus paplionus 0-52 003 0-08 -0-08 Podocarpites reductus -0-18 013 0-29 -017 Palaeoconifereus asecatus -0-02 001 003 0-04 Callialasporites dampieri -0-34 0-24 0-23 -0-37 Callialasporites turbatus -0-23 016 0-26 -018 Callialasporites trilobatus -0-07 -004 0-04 -009 Callialasporites microvelatus -0-25 0-25 0-28 -0-29 Cerebropollenites macroverrucosus -0-24 -0-18 -012 -0-04 Perinopollenites elatoides -0-35 0-18 0-23 -006 Perinopollenites tumulosus -004 -015 -010 0 Microcachrydites antarcticus -001 -002 0-01 0-04 Clavatipollenites - Rhaetic sp. 0-02 003 0-02 006 Densoisporites (Rhaetic spp.) 0-46 0 006 -014 Taeniaesporites rhaeticus 0-27 Oil 0-04 004 Anemiidites echinatus 0-54 -015 0-08 -0-08 Granuloperculatipollis rudis 009 0-13 0-01 -004 52 PALAEONTOLOGY, VOLUME 40 TABLE 2. (cont.) Principal component: 1 2 3 4 Limbosporites lundbladii 025 002 005 008 Ovalipollis ovalis 033 036 003 -010 Rhaetipollis germanicus 037 036 004 -009 Cornutisporites seeburgensis 017 -002 003 0-02 Semiretisporis gutlrae 015 -001 0-04 001 Aratrisporites fimbricatus 003 003 0-02 007 Protohaploxypinus microcorpus 0-07 002 003 008 Proportion of total variance (per cent.) 5-51 3-64 3-62 2-93 own opinion. Vitreisporiles pallidus illustrates some of the problems: Couper (1958) thought it was from a caytonialean; Filatoff (1975) thought it was a cycad spore; while many subsequent palynologists have attributed it to a conifer. INTERPRETATION OF THE POLLEN GROUPINGS There are four reasons for our interpretations of the palynological groupings emerging from the statistical analyses. One reason for our climatic interpretation is that, as the geographical and chronological spread of the evidence increases, the patterns revealed by multivariate statistical analyses rapidly become more general. In Quaternary studies, 'ecological' patterns give way to ‘climatic’ ones when the data from one geographical region are augmented by information scattered across a continent. When the chronological scale is extended to tens of millions of years from tens of thousands of years, in our experience temperature alone dominates as the source of variance in the statistical analyses based on correlation matrices. Oddly enough, evolution seems to contribute relatively little variance in these kinds of analyses, even with dinoflagellates whose fast reproduction and changing cyst-types imply rapid evolution, which would lead one to expect such problems. (Observer-errors in the form of taxonomic inconsistencies between microscopists can be an even more important source of variance. Changing patterns of catchment and transport of pollen can have dramatic effects on pollen spectra; thus the non-trivial principal components typically account for 90-95 per cent, of the total variance in a Pleistocene pollen analysis, while in deep boreholes and other deposits with complex taphonomies the figure falls to 60-70 per cent.) The dominant role of temperature in long- term climatic developments is illustrated in our Tertiary studies (Hubbard et al. 1994) where the pollen and dinoflagellate cyst records parallel the temperature record from oxygen isotope analyses in an unmistakable way. Secondly, and more specifically, the Sun is the Earth’s major heat source, and the Second Law of Thermodynamics ensures that warmth is distributed quite rapidly from the equatorial regions to both poles equally. The distribution of moisture is essentially a secondary process driven by this primary flow of insolation-derived heat. Le Chatelier's Principle similarly demands that concentrations of water vapour tend to equilibrate, but the process is slower and is strongly influenced by local factors, as the world’s deserts show. Low humidity, however, tends to lead to low vegetational cover, and to low pollen production. There are therefore excellent a priori grounds for expecting correlations that are made over long distances (and particularly ones between the southern and northern hemispheres) to reflect temperature changes on a world-wide scale. These considerations, with the uniformitarian principle that the (distant) past behaved in the same way as the relatively recent, and Occam’s entia non sunt multiplicanda praeter necessitatem , led us to expect our results to be interpretable in terms of global temperatures. In fact, although they HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 53 lack subjective appeal, these arguments demand that no other interpretations should be considered unless temperature change can be excluded. If these last two reasons insist that our findings must be interpreted in terms of global temperatures, the specific identifications of our climatic groupings are more ambiguous. Our initial assignation to our palynological groups of climatic associations arose from the recognition from the Australian evidence, that the groupings corresponded quite strongly to phylogenetic structures. Group S I is dominated by conifers, S II by primitive gymnosperms (e.g. Cheirolepidiaceae, cycads), and S III by moss- and fern-like organisms (a similar pattern is detectable in the northern hemisphere results, but is obscured by the complexity and far greater heterogeneity of the raw data). The correlation implied that Group S I was associated with much cooler and drier conditions than the others (since conifers are conspicuously more tolerant of drought and cold than ferns, mosses, and cycads) and that S III was tropical in character. Further evidence for this general interpretation comes from the inverse correlation between the representation of our cold-climate group in the Hasty Bank section (Text-fig. 4) and the absolute concentrations of megafossils recorded by Hill (in Spicer and Hill 1979). At the base of the Hasty Bank section, the absolute concentration of macro-remains was about 80000 per m3. Immediately above this the concentration rose to around 130000 per m3, then it declined irregularly to about 10000 per m3. Other things being equal, this pattern suggests a climatically controlled increase and decrease in primary production, consistent with our climatic interpretations. Given that group S II is strongly represented in the Australian strata the colour of which could suggest either arid or humid tropical conditions (FilatofF 1975) it is conceivable that the attributions of our tropical and intermediate groups might be inverted. Such a re-attribution would pose problems, however. Group N II dominates the Rhaetian pollen spectra at Kendelbach (Morbey 1975) (Text-fig. 7), and the aeolian character of many later Triassic terrestrial sediments has generally been taken to imply an arid climate, not a humid one. It is frustrating that both the northern and southern hemisphere sets of data come from rather similar latitudes, limiting the deductions that can be made on internal evidence. None the less, the climatic interpretations we have placed on our groupings can be seen to show both internal consistency, and consistency with their palaeogeographical contexts. Comparison of the climatic conditions associated with the 15 or so pollen and spore types which occur in both hemispheres reveals that about 40 per cent, belong to the equivalent climatic group (e.g. Gleicheniidites senonicus , Vitreisporites pallidus), and just under half move from N I to S II (e.g. Calamospora mesozoica , Dictyophyllidites harrisii ), consistent with the differences in palaeolatitude. Only two (Calliala- sporites turbatus, Matonisporites crassiangulatus ) shift in the 'wrong direction’. Once the palynological groupings have been related to approximate climatic conditions, then these can be used to make crude climatic reconstructions (Hubbard and Boulter 1983) (see below); Text-figures 4 and 7 show estimates of summer and winter temperatures calculated in this way. CORRELATION AND DATING We have correlated one Yorkshire pollen sequence (Hasty Bank) spanning the Toarcian-Aalenian boundary with the Australian master sequence. Other British sequences correlated with the Australian diagrams identify the position of the Aalenian-Bajocian boundary; and a pair of Bathonian sequences are thought to span the Morrisiceras morrisi - Proceras hodsoni Oppel-zone boundary (Text-fig. 4). Our correlations are consistent with the existing ammonite evidence, but (being based on quantitative analyses) are more accurate and reliable. The absolute chronology of the Jurassic and Triassic periods has long been a matter of controversy. Our objectively based estimates of the relative lengths of the Aalenian, Bajocian and Bathonian stages show that the 'standard’ chronology (Hallam et al. 1985) is much less satisfactory than van Hinte’s chronologies (1976). Our most accurately defined points are in closer agreement with the latter’s preferred chronology, which has therefore been used as the chronological basis of 54 PALAEONTOLOGY, VOLUME 40 Z Z Z TEXT-FIG. 3. For caption see opposite. z our work. Table 5 gives our key datings using the optimum lit between our relative chronology and van Hinte’s. Although the Bajocian-Bathonian boundary is not defined by any of the sequences studied, its position can be estimated with some accuracy. In the Aalenian and earlier Bajocian, the Oppel ammonite zones seem to correspond to warm-to-warm cycles (which would imply the extinction of cold-loving ammonites by warm episodes) whose periodicities averaged 142 million years (Ma)- indicating a date of about 162-7 Ma for the base of the Bathonian. With the Morrisiceras morrisi- Procerites hodsoni boundary apparently fixed by the Gristhorpe and Cambridge sections, one can HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 55 z text-fig. 3. Average Link dendrogram showing the relationships between the northern hemisphere Mesozoic pollen and spores in the 680 samples involved in this study. Sporomorphs of essentially cold environments are designated as Group N I, thermophilous ones are classified as Group N III; and Group N II contains the climatically intermediate taxa. The attributions of the palynomorphs to families are very speculative, but follow the opinions of the European palynologists cited. calculate that in the Bathonian, the average duration of an Oppel-zone was rather shorter (0-80 Ma). Application of this estimate above the fixed Bathonian horizon supports FilatoflT s idea that the sharp warming at about 155 Ma is indeed the Bathonian-Callovian boundary. The accuracy with which correlations may be made using such quantitative evidence sometimes poses problems. The precision of the correlation depends - other things being equal - on the PALAEONTOLOGY. VOLUME 40 56 PALAEONTOLOGY, VOLUME 40 table 3. Simplified enumeration of the rotated principal components yielded by the Australian data. The values tabulated are rounded integer tenfold multiples of the rotated principal component loadings (thus a value of 7 reflects a loading between +0 65 and 0-74): loadings of less than 0-25 (regardless of sign) have been ignored. Rotated principal component: 1 2 3 4 5 6 7 8 9 1011 1213141516171819 20 Slereisporites psilatus Stereisporites antiquasporites Rogalskaisporites cicatricosus Rogalskaisporites canaliculus Polycingulatisporites crenulatus Polycingulatisporites striatus Antulsporites varigranulatus Antulsporites clavus Antulsporites saevus Foveosporites moretonensis Staplinsporites caminus Staplinsporites telatus Staplinsporites mathurii Staplinsporites perforatus Foveosporites multifoveolatus Foveotriletes cf. irregulatus Densoisporites circumundulatus Densoisporites sp. A Densoisporites sp. B Lycopodiacidi tes cerniidites Camarozonosporites ramosus Camarozonosporites clivosus Leptolepidites major Uvaesporites equibossus Uvaesporites crassibalteus Uvaesporites sp. Lophotriletes sp. Apiculatosporites Acanthotriletes levindensis Conbaculatisporites cf. mesozoicus Neoraistrickia taylorii Neoraistrickia pilobaculata Neoraistrickia trichosa Neoraistrickia densata Neoraistrickia truncata Neoraistrickia sp. A Neoraistrickia spp. Lycopodiumsporites rosewoodensis Lycopodiumsporites reticulumsporis Lycopodiumsporites austroclavatis Lycopodiumsporites circolumensis Lycopodiumsporites sp. A Lycopodiumsporites spp. Dictyosporites speciosus Dictyosporites complex Paxillitriletes Pilasporites marcidus Calamospora mesozoica Marattisporites scabratus Todisporites minor Osmundacidites wellmanii Baculatisporites comaumensis Verrucosisporites various Verrucosisporites rugulodomus Rugulatisporites chamiomatus 5 5 3 -4 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 57 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 56 PALAEONTOLOGY, VOLUME 40 TABLE 3. Simplified enumeration of the rotated principal components yielded by the Australian data. The values tabulated are rounded integer tenfold multiples of the rotated principal component loadings (thus a value of 7 reflects a loading between +0-65 and 0-74): loadings of less than 0-25 (regardless of sign) have been ignored. Rotated principal component: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Stereisporites psilatus Stereisporites antiquasporites Rogalskaisporites cicatricosus Rogalskaisporites canaliculus Polycingulatisporiles crenulatus Polycingulatisporites striatus Antulsporites varigranulatus Antulsporites clavus Antulsporites saevus Fo veosporiles moretonensis Staplinsporites caininus Staplinsporites telatus Staplinsporites mathurii Staplinsporites perforatus Foveosporites multifoveolatus Foveotriletes cf. irregulatus Densoisporites circumundulatus Densoisporites sp. A Densoisporites sp. B Lycopodiacidites cerniidites Camarozonosporites ramosus Camarozonosporites cli vosus Leptolepidites major Uvaesporites equibossus U vaesporites crassibalteus Uvaesporites sp. Lopliotriletes sp. Apiculatosporites A can tho triletes le vindensis Conbaculatisporites cf. mesozoicus Neoraistrickia taylorii Neoraistrickia pilobaculata Neoraistrickia trichosa Neoraistrickia densata Neoraistrickia truncata Neoraistrickia sp. A Neoraistrickia spp. Lycopodiumsporites rosewoodensis Lycopodiumsporites reticulumsporis Lycopodiumsporites austroclavatis Lycopodiumsporites circolumensis Lycopodiumsporites sp. A Lycopodiumsporites spp. Dictyosporites speciosus Dictyosporites complex Paxillitriletes Pilasporiles marcidus Calamospora mesozoica Marattisporites scabratus Todisporites minor Osmundacidites wellmanii Baculatisporites comaumensis Verrucosisporites varians Verrucosisporites rugulodomus Rugulatisporites chamiomatus . .7 8 5.5 6 3 3 6 8 . . . . 8 5 .... 3 5 5 . .3.73 6 . . . . . . 8 7 8.3 4.5. .6 7 4 3 3 . . . 7 7 . . . 9 3 4 3 7 4 7 9 3 5 3 6 3 7.. 3 . -4 8 3 5 3 6 3 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 57 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 7 7 3 . . . 3 4 . . . 5 3 7 7 8 3 -3 3 ... 5 9 3 .... 3 ... 4 7 7 6 7 4 3 3 7 4 7 58 PALAEONTOLOGY, VOLUME 40 TABLE 3. (cont.) Rotated principal component: 1 2 3 4 5 6 7 8 9 1011 1213 141516171819 20 Cyathidites australis Cyathidites minor Cibotiumspora juriensis Dictyophyllidites equiexinus Die tyophy llidites liarrisii Matonisporites crassiangulatus Anapiculatisporites clausonensis Foraminisporites tribulosus Gleicheniidites senonicus Duplexsporites problematicus Contignisporites cooksonii Trilobosporites antiquus Ischyosporites cratens Ischyosporites vollcheimeni Ischyosporites marburgensis Klukisporites variegatus Klukisporites neo variegatus Klukisporites lacunus Klukisporites scaberis Lygodioisporites perverrucatus Retusotriletes mesozoicus Granulatisporites sp. Converrucosisporites variverrucatus Cadargasporites baculatus Cadargasporites reticulatus Nevesisporites vallatus Murospora florida Lycopodiacidites asperatus Monolites couperi Reticuloidosporites sp. Obtusisporites varragadensis Obtusisporites cf. canadensis Cycadopites follicularis Vitreisporites pallidus Sulcosaccispora lata Alisporits lowoodensis Alisporites cf. grandis Pinuspollenites parvisaccatus Pinuspollenites globosaccus Podocarpidites ellipticus Podocarpidites cf. verrucosus Podocarpidites cf. multesimus Podosporites castellanosii Araucariacites australis Callialasporites segmentatus Callialasporites minus Callialasporites microvelatus Callialasporites dampieri Callialasporites turbatus Callialasporites trilobatus Classopollis chateaunovi Classipollis simplex Classopollis cf. meyeriana Classopollis anasillos Exesipollenites tumulus Exesipollenites scabratus . -4 3 . -3 9 . 5 6 3 . 6 . . 7 3 7 3 5 9 4 . . 3 . . . 4 . . 3 . 6 . . 3 . . 8 8 9 7 3 3 4 4 8 3 7 6 -4 -3 4 3 7 5 4 7 . . 7 . . 3 3 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 59 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 7 . 5 3 5 3 3 58 PALAEONTOLOGY, VOLUME 40 TABLE 3. ( cont .) Rotated principal component: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Cyathidites australis Cyathidites minor Cibo tiumspora juriensis Dictyophyllidites equiexinus Dictyophyllidites harrisii Matonisporites crassiangulatus A napiculatisporites clausonensis Foraminisporites iribulosus Gleicheniidites senonieus Duplexsporites problematicus Contignisporites cooksonii Trilobosporites antiquus Ischyosporites cratens Ischyosporites vollcheimeni Ischyosporites marburgensis Klukisporites variegatus Klukisporites neovariegatus Klukisporites laeunus Klukisporites scaberis Lygodioisporites perverrucatus Retusotriletes mesozoicus Granulatisporites sp. Converrucosisporites variverrucalus Cadargasporites baculatus Cadargasporites reticulatus Nevesisporites valiants Murospora florida Lycopodiacidites asperatus Monolites couperi Reticuloidosporites sp. Obtusisporites varragadensis Obtusisporites cf. canadensis Cycadopites follicularis Vitreisporites paUidus Sulcosaccispora lata Alisporits lowoodensis Alisporites cf. grandis Pinuspollenites parvisaccatus Pinuspollenites globosaccus Podocarpidites ellipticus Podocarpidites cf. verrucosus Podocarpidites cf. multesimus Podosporites castellanosii Araucariacites australis Callialasporites segmentatus Callialasporites minus Callialasporites micro velatus Callialasporites dampieri Callialasporites turbatus Callialasporites trilob at us Classopollis cliateaunovi Classipollis simplex Classopollis cf. meyeriana Classopollis anasillos Exesipollenites tumulus Exesipollenites scabratus . .3 8 9 7 4. .3. 6 . . 3 . . 8 3 . . 6 3 7 4 3 3 4 4 5 4 7 7 3 3 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 59 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 7 3 3 3 3 3 3 3 7 . . 8 3 . . 3 3 5 8 7 5 3 8 60 PALAEONTOLOGY, VOLUME 40 table 4. Simplified enumeration of the rotated principal components yielded by the European data, treated as in Table 3. Rotated principal component : 1 2 3 4 5 6 7 8 9 10111213141516171819 20 Stereisporites stereoides Stereisporites antiquasporites Sculptisporis seeburgensis Foveasporis irregularis Polycingulatisporites circulus Staplinisporites caminus Densoisporites velatus Neoraistrickia gristliorpensis Lycopodiacidites rugulatus Lycopodiacidites cerniidites Lycopodiacidites rhaeticus Relitriletes pseudoreticulatus Retitriletes austroclavatidites Relitriletes davatoides Foveotriletes microreticulatus Lyeopodiumsporites sp. A Rotverrusporites major Rotverrusporites equatibossus Rotverrusporites bossus Leptolepidites plurituberosus Contignisporites glebulentus Aequitriradites spinulosus Vallizonosporites pseudoalveolatus Cicatricosisporites brevillaesurae Couperisporites complexus Duplexisporites problematicus Heliosporites reissingeri Calamospora mesozoica Marattisporites scabratus Todisporites major + minor Osmundacidites wellmanii Pilosisporites brevipappillosus Concavisporites crassexinus Trilobosporites bernissartensis Concavissimisporites variverrucatus Deltoidospora australia Deltoidospora concavus Deltoidospora neddeni Deltoidospora equiexinus Dictyophyllidites harrisii Camarozonosporites rudis Krauselisporites linearis Endosporites rhysoseus Endosporites jurassicus Ricciisporites tuberculatus Matonisporites crassiangulatus Cibotiumspora sinuata Cibotiumspora tricuspidata Obtusisporis canadensis Gleicheniidites senonicus Ischyosporites lacunus Trachysporites asper Undulatisporites concavus T riancoraesporites ancorae Platyptera trilingua Convolutispora microrugulatus Auritulinasporites scanicus Anapiculatisporites cronaferreus Uvaesporites argentaeformis Nevesisporites bigranulatus HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 61 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 -3 -3 5 60 PALAEONTOLOGY. VOLUME 40 table 4. Simplified enumeration of the rotated principal components yielded by the European data, treated as in Table 3. Rotated principal component : I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Stereisporites stereoides Stereisporites antiquasporites Sculptisporis seeburgensis Foveasporis irregularis Polycingulatisporites circulus Staplinisporites caminus Densoisporites velatus Neoraistrickia gristhorpensis Lycopodiacidites rugulatus Lycopodiacidites cerniidites Lycopodiacidites rhaeticus Retitriletes pseudoreticulatus Retitriletes austroclavatidites Retitriletes clavatoides Foveotriletes microreticulatus Lycopodiumsporites sp. A Rotverrusporites major Rotverrusporites equatibossus Rotverrusporites bossus Leptolepidites plurituberosus Contignisporites glebulentus Aequitriradites spinulosus Vallizonosporites pseudoalveolatus Cicatricosisporites brevillaesurae Couperisporites complexus Duplexisporites problematicus Heliosporites reissingeri Calamospora mesozoica Maratlisporites scabratus Todisporites major + minor Osmundacidites wellmanii Pilosisporites brevipappillosus Concavisporites crassexinus Trilobosporites bernissartensis Concavissimisporites variverrucatus Deltoidospora australia Deltoidospora concavus Deltoidospora neddeni Deltoidospora equiexinus Dictyophyllidites harrisii Camarozonosporiles rudis Krauselisporites linearis Endosporites rhysoseus Endosporites jurassicus Ricciisporites tuberculatus M atonisporites crassiangulatus Cibotiumspora sinuata Cibotiumspora tricuspidata Obtusisporis canadensis Gleiclieniidites senonicus Ischyosporites lacunus Trachysporites asper Undulatisporites concavus Triancoraesporites ancorae Platyptera trilingua Con volutispora microrugulatus Auritulinasporites scanicus Anapiculatisporites cronaferreus U vaesporites argen taeformis N e vesispori tes bigranulatus 3 3 7 3 7 4 7 6 5 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 61 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 3 5 3 -3 3 5 5 5 9 5 3 4 3 5 6 62 PALAEONTOLOGY, VOLUME 40 TABLE 4. ( cont .) Rotated principal component: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Laevigatosporites couperi Lygodioisporites perverrucatus Plicatella tricornitata Reticulisporites venulosus Cicatricosisporites dorogensis Murospora spp. Clavifera triplex Zebrasporites spp. Classopollis torosus Classopollis reclusus Classopollis meyeriana Inaperturopollenites australis Chasmatosporites spp. Schizosporis spriggi Peltandripites tener Undulatasporites callosus Naiaditaspora anglica Cvcadopites carpentieri Cycadopites subgranulosus Cvcadopites minimus C la vatipollenites hughe sii Eucommiidites troedssonii Phyllocladites microreticulatus Walchites sp. Vitreisporites pallidus Spheripollenites scabratus Spheripollenites psilatus Pinuspollenites minimus A lisporites dunrobinensis Alisporites microsaccus Alisporites thomassii Alisporites robustus Pityosporites similis Pityosporites microalatus Pityosporites scaurus Polonisaccus ferrugineus Parvisaccites enigmatus Platysaccus paplionus Podocarpites reductus Palaeoconiferus asecatus Callialasporites dampieri Callialasporites turbatus Callialasporites trilobatus Callialasporites microvelatus Cerebropollenites macroverrucosus Perinopollenites elatoides Perinopollenites tumulosus M icrocachrydites antarcticus Clavatipollenites Rhaetic sp. Densoisporites (Rhaetic spp.) Taeniaesporites rhaeticus Anemiidites echinatus Granuloperculatipollis rudis Limbosporites lundbladii Ovalipollis ovalis Rhaetipollis germanicus Cornutisporites seeburgensis Semiretisporis guthae Aratrisporites fimbricatus Protohaploxypinus microcorpus 7 8 5 -4 3 3 5 -3 HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 63 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 8 7 6 7 6 4 7 7 3 3 6 3 62 PALAEONTOLOGY. VOLUME 40 TABLE 4. (com.) Rotated principal component: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Laevigaiosporites couperi Lygodioisporites perverrucatus Plicaiella tricornitata Reticulisporites venulosus Cicairicosisporites dorogensis Murospora spp. Clavifera triplex Zebrasporites spp. Classopollis torosus Classopollis redusus Classopollis meyeriana InaperturopoUenites australis Chasmatosporites spp. Schizosporis spriggi Peltandripites tener Undulatasporites callosus Naiaditaspora anglica Cvcadopites carpentieri Cycadopites subgranulosus Cycadopites minimus Clavatipollenites hughesii Eucommiidites troedssonii Phyllocladites microreticulatus Walchites sp. Vitreisporites pallidus Splieripollenites scabratus Spheripollenites psilatus Pinuspollenites minimus A lisporites dunrobinensis Alisporites microsaccus Alisporites thomassii Alisporites robust us Pityosporites similis Pityosporites microalatus Pityosporites scaurus Polonisaccus ferrugineus Parvisaecites enigmatus Platysaccus paplionus Podocarpites reductus Palaeoconiferus asecatus Callialasporites dampieri Callialasporites turbatus Callialasporites trilobatus Callialasporites microvelalus Cerebropollenites macroverrucosus Perinopollenites elatoides Perinopollenites tumulosus M icrocachrydites antarcticus Clavatipollenites - Rhaetic sp. Densoisporites (Rhaetic spp.) Taeniaesporites rhaeticus Anemiidites echinatus Granuloperculatipollis rudis Limbosporites lundbladii Ovalipollis oval is Rhaetipollis germanicus Cornutisporites seeburgensis Semiretisporis guthae A ratrisporites fimbricatus Proioliaplox) pin us m icrocorpus HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 63 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 64 PALAEONTOLOGY, VOLUME 40 text-fig. 4. Australian and European earlier Jurassic pollen diagrams. The left-hand part of the figure shows composite plots of climatic reconstructions (winter and summer temperatures) and pollen diagrams for the earlier Juiassic in Western Australia. The left-hand hatched areas in the pollen diagrams are the warmth-loving species (S III), the right-hand hatched areas the types indicative of cool climates (S I), and the blank areas HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 65 Ma I . I , . . . I , . . , I . . . I . . . . L text-fig. 5. Age-verms-depth plots for the Cockburn 1, Badaminna 1 and Gun Island 1 boreholes, showing the correlative points that provide the justification for the conflations in Text-fig. 4. The crucial points in the Badaminna 1 borehole are indicated with circles, those at Cockburn 1 with triangles, those at Gun Island 1 with diamonds. Other sample points are shown by cross bars. table 5. Selected chronological events identified in the pollen diagrams, the dates of which can therefore be related to our modified version of van Hinte’s chronology (1976). Stratigraphical event Time Morrisiceras morrisi-Procerites hodsoni boundary 158-62 Ma (Lower-Middle Bathonian boundary) Bathonian-Bajocian boundary 162-7 Ma ? Stephanodiscus hwnpliriesicmum-Eniileia 165-89 Ma (Otoites) sauzei boundary Millepore Bed 168-68 Ma Bajocian-Aalenian boundary 170-15 Ma Eller Beck Bed 170-5 Ma Aalenian-Toarcian boundary 173-5 Ma between depict the climatically intermediate taxa (S II). At the right hand side of the figure, some of the correlative pollen diagrams from north-west Europe and Australia are shown. The vertical scale is in millions of years, using the chronology of van Hinte (1976). The Toarcian-Aalenian boundary is arrowed. (The elements of the composite Australian pollen diagram are identified by marginal marks. Spectra from the Cockburn 1 borehole are indicated by long bars, those from Gun Island 1 by short bars, and the Badaminna spectra are unmarked.) 66 PALAEONTOLOGY, VOLUME 40 172.5 - 173.0 - 173.5 - kfaety Bank text-fig. 6. Toarcian-Aalenian pollen diagrams from Yorkshire, showing the various positions at which the boundary (arrowed) has been placed by Muir (1964) at Hasty Bank and Wilkinson (1978) at Blea Wyke. Blea Wyke Conventions as for Text-fig. 4. O Ma stratigraphical resolution of the curves being matched. The more detailed sections of the Australian master-sequence clearly have a resolution of a few hundreds of thousands of years, and some of the European profiles are sampled at intervals of less than a hundred thousand years. Since it is the patterns that are being matched and not individual spectra, the precision is rather greater than either of these figures suggests. At this level of accuracy, inconsistencies in the positions of the stage boundaries can be perceptible. For instance, the Toarcian-Aalenian boundary was placed by Muir (1964) in the warming period preceding the warm episode, at about 173-59 Ma. However, Couper’s (1958) Whitby sequence and Wilkinson’s (1978) analyses of Blea Wyke Point seem to place the boundary about 350000 years later, at the end of the warm oscillation (Text-fig. 6). Clearly an examination is needed of the definition of the Toarcian-Aalenian boundary and how it relates to these sequences. Other sequences are too closely sampled to be correctable at present ; for instance, the 13 samples analysed by Muir from the 0-9 m Low Beast Cliff section record a sharp N III peak followed by an equally sharp minimum in the N I group. Whether these warm episodes are 20000, 40000, or 100000 years apart it is impossible to say, as the pattern cannot be matched in any of the closely sampled European Aalenian sequences, and the resolution of the Australian sequences is far too coarse. THE TRIASSIC-JURASSIC BOUNDARY Text-figure 7 shows the ecological and climatic developments in north central Pangaea in the final stage of the Triassic Period and at the boundary with the Jurassic. Given the absence of any convincing overlap between the sequences used in Text-figure 4 and these Rhaetian-Hettangian ones, the standard chronology (Forster and Warrington 1985) has been used. It was assumed that the Kendelbach deposits (Morbey 1975) spanned the entire Rhaetian stage. The most important feature is the extremely dramatic cooling episode immediately below the Triassic -Jurassic boundary. It appears to have been considerably more intense (if slightly more gradual) than the global cooling that occurred just above the Eocene-Oligocene boundary (Shackleton 1986; Hubbard et al. 1994), when a global temperature drop of 6 °C took place. Although traditional ideas about global extinctions and climatic changes at the Triassic-Jurassic boundary have been questioned (Weems 1992), the graphs in Text-figure 7 provide direct evidence of major ecological crisis at this time - whether or not this caused mass extinctions. Our evidence indicates that the earliest Jurassic marked a change to persistently and consistently colder conditions. Comparable climatic reconstructions in the Tertiary are demonstrably inaccurate for absolute values of temperature, but give realistic pictures of relative changes (Hubbard and Boulter 1983). In the circumstances, we have made no attempt to relate our climatic curve to exact temperatures. None the less it is fair to observe that, while the Aalenian-Bajocian cold episodes nominally reflect occasional mild winter frosts, the conditions indicated at the Triassic-Jurassic boundary are of consistently severe conditions for several hundred thousand years. It is not clear HUBBARD AND BOULTER: MID MESOZOIC FLORAS AND CLIMATES 67 KendeiSadi Valfiall IfiVernock- Lfirne. O T o (N 317 Grossly thickened 76 PALAEONTOLOGY, VOLUME 40 150 7 A Pr Ar My Pt Lu He An Thickness (^m) Clades text-fig. 2. a, distribution of periostracal thickness for the entire data set. B, histogram demonstrating the bias of the data set; dashed bars show the percentage of extant bivalve families which fall into each clade and solid bars the percentage of taxa from this study belonging to each. An = anomalodesmatans; Ar = arcoids; He = heteroconchs; Lu = lucinoids; My = mytiloids; Pr = protobranchs and Pt = pteriomorphs. Taxonomic variation Text-figure 2a shows the distribution of periostracal thickness observed for the entire data set. Such a plot cannot be considered as representative of the variation in this parameter for the class Bivalvia, as it is undoubtedly biased by the specifics of the set, for example by the over-representation of pteriomorphs and mytiloids and the comparatively low number of anomalodesmatan, lucinoid and arcoid species measured (Text-fig. 2b). Only the lucinoids were seriously under-sampled; of the 13 extant families only seven were investigated. Such incomplete coverage was, in part, due to the small size and fragility of members of the missing families, and their paucity in museum collections. Both of these reasons are in turn the result of their mainly cryptobyssate habits (Yonge and Thompson 1976). Histograms showing the range of periostracal thickness recorded in each clade are given in Text-figure 3, and are discussed below, together with statistical analyses of their significance. Protobranchs. Members of the protobranch clade are characterized by a moderate to thick, obvious and persistent periostracum (Text-fig. 3a). The thickness recorded for these taxa ranges between 3 pi n for Leda minuta and Nucula nucleus to over 100 /mi for species of Solemya. Measured S. borealis attain a periostracal thickness of 100 //m, whilst Beedham and Owen (1965) reported that of S. parkinsoni as being 140 pm. In the solemyoids the periostracum extends ventrally some way beyond the edge of the calcareous shell to form a flexible, radially pleated flange, which Beedham and Owen (1965) showed to be in intimate contact with the mantle epithelium, being the site of orbicular muscle insertions. In all protobranchs examined, the periostracum was smooth with no external ornament or indication of internal structures. Arcoids. Arcoids appear to possess a very thick and persistent periostracum (Text-fig. 3b). However, in many genera, e.g. Limopsis and Glycymeris, the layer is densely covered by conspicuous hairs, the presence of which often makes the precise thickness of the sheet difficult to ascertain. In fact, when HARPER: MOLLUSCAN PERIOSTRACUM 77 text-fig. 3. Histograms showing the range of periostracal thickness recorded for each clade. a, protobranchs; B, arcoids; c, mytiloids; d, pteriomorphs; E, lucinoids; F, heteroconchs; G, anomalodesmatans. one subtracts the thickness of the hairy sub-layer, the basal sheet, to which the hairs are attached, is only moderately thick. For example, Limopsis marionensis has a very conspicuous shaggy pile attached to a sheet of a mere 5 //m thickness. Hair formation in arcoids has been described by Waller (1980) who argued that they are formed during maturation of the periostracal sheet, and not under direct mantle control. Mytiloids. This clade comprises one extant family, the Mytilidae, but contains a great diversity of genera and species. In general the mussels possess a particularly thick and persistent periostracum, and have the thickest periostraca recorded herein (Text-fig. 3c). The recorded range is from 5 pm for Ciboticola lunata (a questionable mytilid; Moore 1969) to 428 pm for Musculus laevigatus. The median value for the data sub-set was 30 //m, with seven taxa registering values of over 100 /mi. The mytiloids also display a range of periostracal structures. As noted by Dunachie (1963) the periostracum of Mytilus edulis is tri-layered (Text-fig. 4a), the central layer often being vacuolated. This layer is not continuous over the entire periostracal area, and its adaptive significance is not clear. Such vacuolation has not been observed in the periostracum of other mussels studied. Hairs were encountered in a number of taxa, e.g. Modiolus modiolus , Modiolus capax and Trichomya hirsutus. They are not simple projections but display a variety of flattened, serrated and palmated morphologies as illustrated by Ockelmann (1983). The functional morphology of these structures was discussed by Bottjer and Carter (1980) who suggested a variety of functions, for example supplementation of shell ornamentation for stabilization, extension of mantle sensors and deterrence of settling by fouling epibionts. These authors considered that the hairs are produced by the outer middle mantle fold, but Ockelmann (1983) observed their formation in Modiolus and juvenile Mytilus on areas of the shell away from the valve edges (and hence away from mantle 78 PALAEONTOLOGY, VOLUME 40 text-fig. 4, a, Mytilus edulis (mytiloid); SM X. 27504; Oban, UK; note the vacuous middle layer; x 1500. b. Pinna saccata (pteriomorph); SM X. 27505; New Caledonia; note the fine, wrinkled periostracal sheet emerging from beside the ciliated surface of the middle mantle fold; x 500. c, Cardita affirms (heteroconch); SM X. 27506; awn-like processes arising on the outer surface of the fine periostracal sheet; x 500. D, Lyonsia norwegica (anomalodesmatan); SM X. 27507; Northumberland, UK; note adherent sand grains (S) on outer surface of periostracum; x 1500. influence). He concluded that they are, like the byssus, produced by the foot and further noted that taxa with the greatest development of these hairs are also those with large anterior byssal gland complexes. Pteriomorphs. Extremely thin (< 1 /mr) periostracal sheets characterize virtually all of the pteriomorph taxa measured (Text-figs 3d, 4b). Fourteen of the 15 extant families of this clade possess such a gossamer-thin sheet that is not easily perceived on the external surface of the shell, and indeed seldom persists past the valve margins. Only in the Anomiidae were thicker periostraca recorded, for example Anomia ephippium (10 /mr), A. archaeus (8 pm) and Monia squamosa (10 /mi). Lucinoids. Members of this clade possess a moderately thick periostracum (Text-fig. 3e), the highest values recorded being 10 /mi for Diplodonta diplodonta and Myrtea botanica. The periostracum is often conspicuous as a straw-yellow, varnish-like coating to the shell, and is reasonably persistent. Heteroconchs. The heteroconch clade is the largest considered here, with 36 extant families, and displays the greatest variation in periostracal thickness (Text-fig. 3f). Values as thin as I /mi (or less) have been recorded in I 5 taxa, whilst the thickest periostracum measured was that of Trapezium sublaevigatum at 1 10 pm. Despite the large range for the clade, that within constituent superfamilies is much narrower, for example Arcticoidea 70-1 10 /mi, Chamoidea 1-2 pm, most Cardioidea HARPER: MOLLUSCAN PERIOSTRACUM 79 1-2 //m and Solenoidea 10-50 //m. There is also great variation in periostracal structures and ornaments within this clade. Many taxa bear smooth, apparently featureless, periostracal sheets whilst others show a range of ornaments and structures. The external surface of the periostracum of members of Astarte shows a reticulate ornament (see Saleuddin 1974, figs 16-17), whilst several of the carditids (e.g. Cardita affinis, see Text-fig. 4c) have hairy periostraca. Anomalodesmatans. These possess a moderate to thick, persistent periostracum (Text-fig. 3g). The finest measured was 2 /mi for Jounettia cumingi , whilst the thickest recorded belonged to Lyonsia norwegica at 80 /mi. Members of the clade also display a number of interesting periostracal features, for example the development of calcareous elements within the organic periostracum (Carter and Aller 1975). Aller (1974) described how calcareous spicules, manufactured by the outer mantle fold of Laternula flexuosa , are incorporated into the periostracal sheet where they may provide stabilization. Carter (1978) described similar spicules in the periostracum of the boring gastrochaenid Spengleria rostrata, which he considered may aid the boring process. Other anomalodesmatans, including many of the Pandoroidea, appear to be characterized by having a ‘sticky’, semi-fluid outer layer to the periostracum. In the lyonsiids, sand grains and other debris adhere to this mucoid layer (Text-fig. 4d), which Prezant (1981) believed may camouflage and protect the shell, or assist with stabilization within a shifting substratum. He described the presence of arenophilic glands within the outer mantle fold which, in Lyonsia , he suggested secreted the mucoid into the periostracal groove on to which the rest of the periostracum is then secreted, whilst in Entodesma he suggested that the glands are positioned more distally and secretions pass through the periostracum, perhaps by localized dissolution of the sheet. However, Morton (1987a) has shown that such glands in members of the Thracioidea, Pholadomyiodea and Clavagelloidea are located in the middle mantle folds where they empty on to the (eventual) outside of the newly formed periostracum. Analysis. The Mann-Whitney test was used to test for significant differences in location between pairs of clades to ascertain whether apparent differences were valid. The results of these tests are table 2. Results of the Mann-Whitney test for the significance in location between pairs of clades. In the upper right hand portion of the table significant differences at the 5 per cent, level are marked + , whilst non- significant differences are marked x . The calculated percentage significances are given in the lower left part of the table. An Ar Het Luc Myt Pr Pter An + + + + + + Are 5% — + + X X + Het 0-25 % 0-05 % — X + + + Luc 0-08 % 001 % 97-5% — + + + Myt 0% 90 % 0% 0% — X + Pr 1-4% 89-9 % 001% 0% 7-9 % — + Pter 0% 0% 0% 0% 0% 0% — shown in Table 2. At the 5 per cent, level most were significant except the following pairs: heteroconch and lucinoid, arcoid and protobranch, arcoid and mytiloids and protobranch and mytiloids. Using a Wilcoxon test, 95 per cent, confidence intervals were established for the median of each clade (Table 3). These show clearly that the anomalodesmatan, heteroconch, pteriomorph and lucinoid clades display relatively small ranges of periostracal thickness, whilst in the others the range is far greater. It is also clear that although there is a great deal of similarity in the thicknesses shown by members of the arcoid, mytiloid and protobranch clades, the others are more distinct, albeit with some overlap. 80 PALAEONTOLOGY, VOLUME 40 table 3. Results of the Wilcoxon test to establish the 95 per cent, confidence interval for the median value of each clade. Clade 95 per cent, confidence intervals (/mr) Anomalodesmatans 7-5-12-5 Arcoids 10-0-52-5 Heteroconchs 2-2— 8-5 Lucinoids 3-5-6-0 Mytiloids 26-5^15-0 Protobranchs 11-5-54-0 Pteriomorphs 0-5-0- 5 Anomalodesmatans Heteroconchs Lucinoids Pteriomorphs Mytiloids Arcoids Protobranchs CD O CO o CO m A m v ^ ^ Thickness ^ ^ i ^ ° Classes (pm) CO text-fig. 5. Histograms showing the periostracal thickness recorded for different life habits, a, byssally attached; B, burrowers; c, cementers; D, borers; e, free living (including nestlers and recliners). HARPER: MOLLUSCAN PERIOSTRACUM Life habits Text-figure 5 shows the distribution of periostracal thickness in the exponents of various life habits. The byssate and burrowing habits are exploited by a number of bivalves, not limited to those with a specific periostracal value. It is, however, worth noting that those taxa which have acquired convergently the ability to burrow to great depths, such as Tagelus (Tellinoidea), Solemya (Solemyoidea), and Solen , Pharella and Ensis (Solenoidea) are characterized by thick periostraca. A thick periostracum is shared by those which are active deep burrowers and those which live entombed at depth. Of the more specialized habits, cementers and free-living bivalves appear to be dominated largely by taxa possessing an ultra-thin periostracum, whilst the borers belong to clades with at least moderate development of the periostracal sheet. Possible adaptive significance of these findings are discussed below. Analysis. As above, the Mann-Whitney test was used to test for the significance of differences in location between each pair of life habit groupings. These results are shown in Table 4. At the 5 per table 4. Results of the Mann-Whitney test for the significance in location between pairs of life habit groups. In the upper right hand portion of the table significant differences at the 5 per cent, level are marked +, whilst non-significant differences are marked x . The calculated percentage significances are given in the lower left part of the table. Borer Byssate Cementer Burrower Borer X + + Byssate 16% — + X Cementer 0% 0% — + Burrower 006% 21 % 0% — cent, level there were significant differences for all pairs except two: borers and byssate, and byssate and burrowers. Table 5 shows the 95 per cent, confidence interval ranges of periostracal thickness for each life habit calculated by the Wilcoxon test. The wide range of values displayed by byssate taxa overlaps with those of the borers and burrowers, although the latter two can be distinguished. The narrow ranges shown by cementers and free-living taxa are well separated from the other life habit groups, but are not distinguishable from one another. Relationship to ornament Bivalve taxa with pronounced radial ornament, and the ‘wrinkled’ shells of Mya truncata show no evidence to support the second model of ornament formation. In all cases examined, the periostracal sheet maintained even thickness over the ridges and troughs. There are some instances where bivalves have a very fine scale of surface ornamentation which does appear to support Model 2, for example micro-tubercles (a few micrometres in diameter) on the surface of neotrigoniids (Taylor et al. 1969) and Myochama (pers. obs.) fit into corresponding depressions on the inner surface of their periostraca. However, as shown by Taylor et ai (1969), these depressions are actually caused by the growth of prismatic crystals and post-date the formation of the periostracal sheet. Nicol (1965) noted that the families which make up the clades here considered as protobranchs, arcoids, mytiloids, anomalodesmatans and most of the lucinoids never bear spines. As noted above these are clades with at least moderately thick periostraca. By contrast, the pteriomorphs, with their mostly ultra-thin periostraca, contain some of the most extravagantly ornamented families (e.g. Spondylidae, Ostreidae and Pectinidae). The heteroconch clade contains both virtually unorna- 82 PALAEONTOLOGY, VOLUME 40 Life habits 95 per cent. confidence intervals Borers 11-5—19-5 Byssate 10-25 Cementer 0-5-0- 5 Burrower 6-5-10 Free living 0-5-0- 5 table 5. Results of the Wilcoxon test to establish the 95 per cent, confidence interval for the median value of each life habit group. mented (e.g. solenoids, most venerids and mactrids) and very spiny families (chamids, cardiids and some venerids (e.g. Pitar and Chione)). INTERPRETATIONS In this survey it has been shown that although there is a great variation of periostracal thickness within the Bivalvia, specific ciades, life habit groups and styles of ornamentation are characterized by much narrower ranges. Is there any evolutionary significance to these observations? In his review of Ordovician bivalves Pojeta (1978) considered them to be predominantly either shallow non-siphonate burrowers or forms that were byssally attached either within (endobyssate) or on (epibyssate) the sediment. Both of these life habits may be considered as primitive within the Bivalvia, from which all the other life habits were ultimately derived. Several authors have suggested that many of the specialized life habits have a defensive value and appear to have evolved chiefly after, and in direct response to, the increase in predation pressure at the beginning of the Mesozoic (Vermeij 1987; Harper and Skelton 1993 b). They observed that one of the most interesting aspects of the adaptive radiations of the bivalves is identifying the constraints and preadaptations which have determined the pathways taken by various ciades. Is it possible that the form, in particular the thickness, of the periostracum may have had an important influence? Shallow burrowing and byssate bivalves show a wide range of periostracal thickness and it is difficult to argue that there is any particular primary advantage to any of these, although there are secondary advantages, as discussed above. Patterns only develop when considering the more specialized life habits, and where the acquisition of new habits has been polyphyletic it may be possible to test whether certain habits are associated with particular periostracal characters. Particularly thin periostracal sheets appear to correlate with the cementing and free-living modes of life. The cementing bivalves examined include members of each of the nine ciades of extant marine forms of these (Harper 1991). All but two of these ciades are characterized by the possession of an ultra-thin periostracum. Harper (1992) considered that the ability of these cementers to construct their shells in extremely close proximity to the micro-topography of the substratum is vital to their attachment. This by necessity means that they have a very thin periostracum, and Harper ( 1 99 1, p. 45) noted that ‘ no periostracum can follow substratal irregularities whose radii of curvature are less than twice the periostracal thickness'. Even within the freshwater cementers, Gregoire (1974) noted that although the periostracum of the unionid Etheria is thick over the non-cementing part of the shell, over the attachment scar it is considerably thinned. The two ciades of cementing bivalves which do have more substantial periostraca, Cleidothaerus and Myochama are anomalodesmatans. Morton (1974) suggested that Cleidothaerus cements by means of a ‘sticky’ outer periostracal layer, presumably analogous to the secretions of the arenophilic glands of other pandoroids (Prezant 1981 ; Morton 1987). If this is correct, the problem of creating close proximity between the bivalve and the substratum is solved by the fluid nature of this outer layer. HARPER: MOLLUSCAN PERIOSTRACUM 83 The possession of an ultra-thin periostracum by free-living bivalves is considered to be less significant for two reasons. First, in contrast with the other life habits recognized here, all free-living bivalves belong to a single major clade, the pteriomorphs. Although undoubtedly polyphyletic within that clade, virtually all pteriomorphs measured had ultra-thin periostraca and hence constancy in this autecological category can be assumed to result primarily from their phylogenetic legacy. Any putative advantages of possessing such a thin periostracum, for example as suggested by Moore and Trueman (1971) in the reduction of drag in swimming scallops, are likely to be fortuitous secondary benefits shared with non-swimming members of the same family. Second, this life habit represents a ‘mixed bag’ of occasional swimmers (e.g. some pectinids and limids), recliners (e.g. most pectinids, gryphaeid oysters and placunids) and even free crawlers, in the case of the anomiid Enigmonia engimatica (Yonge 1977). There is no immediately obvious reason why a pleurothetic mussel (with its attendant thick periostracum), if it were to exist, should not succeed in a ‘free-living’ life habit. Particularly thick periostraca appear to be associated with those bivalves which bore chemically into hard substrates and those which burrow deeply within the sediment. Taylor and Kennedy (1969) first noted that borers generally possess thick periostraca and suggested that it protected them from their own boring secretions. It is now confirmed that each of the seven clades of extant boring bivalves recognized by Vermeij (1987) does indeed possess a thickened periostracum (although I was unable to obtain suitable material of the boring arcoid Litharca for the quantitative survey, rather worn valves of L. saxicava , from the collections of The Natural Flistory Museum (London), confirmed the presence of a persistent periostracum). It has been demonstrated that members of each of these clades, with the exception of the pholads, use acidic secretions, produced from glands in either the middle or inner mantle fold, to assist, at least partly, with the boring process (Morton 1990). It seems likely, therefore, that the possession of a thick periostracum is preadaptive for the life habit. Suggestions for the adaptive significance of the thick periostracum in deep burrowing clades are rather more problematical. It may offer the shells of those taxa which burrow actively some protection from sediment scour, or, since the interstitial water in many sediments is undersaturated with respect to calcium carbonate (see Davies et al. 1989), protection against shell dissolution. Therefore, it seems likely that the possession of specific types of periostraca has been preadaptive in the evolution of many specialized life habits and that exponents of these habits have evolved from either shallow-burrowing or byssate taxa. The form of the periostracum in these primitive stocks influenced the pathways used by specific clades during the radiation. Clearly, periostracal traits are not the sole factors involved in the acquisition of these specialized habits; cementation requires also the assumption of a pleurothetic habit and the production of a suitable cement (Harper 1992), while boring requires the evolution of acid-secreting glands (Morton 1990) and deep burrowing the development of a powerful foot and extensive mantle fusion to allow siphon formation (Stanley 1968). The results of this survey also appear to show that surface ornamentation is produced by folding a periostracal sheet of constant thickness, and that the finer the sheet the finer the ornament attainable. There is a simple analogy of attempting origami with cardboard instead of thin paper. Interestingly, Checa (1995) recently published a survey of micro-ornament in ammonoids in which he attributed their formation to folding of the periostracum. The inability of a periostracal sheet of a given thickness to describe an ornament more intricate than a certain scale can be considered in the light of three limiting models: (1) a space-limited periostracum which is highly flexible and limited only by the need to fit physically into the space between the elements of ornament; (2) an energy-limited periostracum where, if too much energy is stored by bending of the sheet, it will pull free of the shell surface; and (3) a strain-limited periostracum where if the sheet is bent too sharply, it will crack. It is tempting to attempt to derive a standard equation whereby it is possible to predict the minimum scale of ornamentation that a periostracum of a given thickness could produce (and which conversely could be used perhaps to calculate the periostracal thickness for fossil taxa). Such an 84 PALAEONTOLOGY, VOLUME 40 equation would, however, depend upon the constant elasticity of the periostracal sheet, and it seems likely that periostraca composed of heterogenous layers, for example the vacuous central layer of the periostracum of Mytilus edulis , or those which appear to possess more fibrous layers, would have a different flexibility from those with a more homogenous structure. Again the periostracum may not be held solely responsible for the evolution of shell ornament. Certainly, the very small scale tubercles on the external surfaces of trigoniids are, as noted previously, the result of individual crystals standing proud. Waller (1972) described how, amongst the pectiniform bivalves (all of which have ultra-thin periostraca), those with outer calcareous shell layers of foliated calcite (e.g. Pectinidae) are able to form shells with sharper radial corrugations and projecting spines and squamae than the generally smoother propeamussids with their outer prismatic shell layers. He attributed this to the smaller size of the microstructural elements which make up foliae rather than prisms, thus enabling the former to take up finer surface ornamentation. This idea is persuasive but is not perhaps the whole story, as many of the pinnids bear intricate hyote spines despite having an outermost prismatic shell layer. That the microstructural unit does not necessarily define the minimum scale of the ornament of bivalves with an outer prismatic shell layer was shown by Carriker et al. (1980), who illustrated the external surfaces of prisms in modern oysters which show distinct keels and furrows. This observation has been repeated for the outer surfaces of Jurassic gryphaeid oysters (Todd and Harper, unpublished data). In these cases the delicate keels and intervening furrows must be produced by minute deflections of the ultra-thin periostracal sheet on to which the prism was seeded. There is a number of advantages that accrue to bivalves able to produce intricate shell ornamentation. Logan (1974) considered a number of functions for the spiny outgrowths of spondylids, which might easily be applied to similar structures in other epifaunal taxa. These include defence (either directly or by promoting camouflaging epibionts), assistance with attachment and stability and protection from fouling. Similarly infaunal bivalves may be shown to gain defensively (Carter 1967) or from stabilization within the sediment (Savazzi 1985). It seems likely, therefore, that the possession of a thick, inflexible periostracum may have been an important constraint in preventing certain bivalve groups from exploiting these morphological benefits. Thomas (1978) regarded the arcoids as having been limited in their evolutionary potential by the relative inflexibility of key elements of their body plan such as shell microstructure, a weak ligamenture and dentition. To this list may be added their thick periostraca and it seems likely that the mussels too have been constrained from great morphological diversity by this trait. No doubt both epifaunal mytiloids and arcoids would benefit in their exposed life habits from the ability to form spines and flanges but they have been unable to produce any more elaborate shell ornament than the coarsest of ribbing (e.g. Geukensia and Septifer). Such disadvantages may be mitigated, at least in part, by the evolution of periostracal hairs, which may be shown to have some defensive and stabilizing value (Bottjer and Carter 1980; Wright and Francis 1984). Similarly the inability to form intricate micro-ornament may be solved, again partially, by the evolution of prefabricated calcareous bodies which can then be incorporated into the periostracum, as described by Aller (1974) for the anomalodesmatan Laternula. THE FOSSIL RECORD AND THE EVOLUTION OF THE PERIOSTRACUM If the form of the bivalve periostracum has played a vital role in their evolution, it would be interesting to know how it has altered over geological time. It might then be possible to answer questions such as whether certain traits are preadaptive or adaptive to certain habits or morphologies. If we accept that the bivalves are a monophyletic group then we must presumably accept that the great variety of periostraca observed in modern representatives is a result of evolution, influenced by different selection pressures away from a primitive periostracum. Presumably thickening and thinning may be achieved either by addition or subtraction of periostracal layers or by varying the thickness of individual layers. But what was the nature of the primitive HARPER: MOLLUSCAN PERIOSTRACUM 85 periostracum? Clark (1976) considered that this was a question that we could never answer but suggested that the ultra-thin periostraca of oysters and scallops are just as derived as the very thickened sheets of mytiloids. Given the apparent distribution of periostracal thickness within the class (Text-fig. 2a), and the characteristics of those bivalves, such as the protobranchs, which are considered to show a number of primitive characteristics (Morton 1996), it seems likely that the primitive periostracum was moderately thick. Direct evidence, however, is difficult to muster because, as noted above, the preservation potential of periostracum is very low. Nevertheless, there is a small number of instances where periostracal preservation has been reported (see Table 6) and table 6. Instances where periostracum preservation has been recorded. § = Crampton (1990); | = Hudson (1968); * = Khz (1972). Taxon Clade Thickness (//m) Age Pholadidea wiffenae § anomalodesmatan > 5 late Cretaceous Praemytilus strathairdensis f mytiloid 5-15 mid Jurassic Cardiola alata* arcoid 30-60 late Silurian Cardiola ti.x* arcoid 30-60 late Silurian Dr T. J. Palmer has pointed out to me that Pojeta (1978, pi. 15, fig. 6) figured a specimen of the mytiloid Modilopsis cf. modiolaris which has a substantial outer layer which could be interpreted as periostracum. It is not surprising that the rare cases where bivalves have been preserved with their periostracum are those that belong to clades whose modern representatives have particularly thick periostraca. Comparison of the figures given in Table 6 with the Recent data set show that each falls within the modern range of their respective clades. As noted by Harper and Skelton (1993a), the value for the Jurassic mytiloid is lower than that recorded for most extant mussels, but we cannot tell if these periostraca were preserved intact. Further cases are required before it would be possible to test whether there has been a post-Jurassic selective thickening of the mussel periostracum (perhaps in response to the radiation of boring gastropod predators; Harper and Skelton 19936). Undoubtedly the preservation of fossil periostraca is biased towards those taxa with thicker periostraca] sheets, and in any case is an exceptional event. A possible avenue for the study of periostracum evolution may be afforded by the examination of ornamentation. Very few Palaeozoic bivalves bear intricate ornamentation and Vermeij (1987) suggested that the first spiny bivalves appeared in the Carboniferous. Although it might be argued that this early lack of ornamentation may be due to the lack of an appropriate extrinsic cue, most obviously intense predation pressure, which would favour its evolution (see Vermeij 1987 ; Harper and Skelton 19936), it might equally be plausible to suggest that it was an intrinsic constraint, i.e. possession of an at least moderately thick periostracum, that prevented it. CONCLUSIONS AND IMPLICATIONS FOR OTHER GROUPS Periostracal thickness has been shown to be an extremely variable character amongst members of the Bivalvia but within specific clades the range of variation is much narrower. It is suggested that difference in periostracal thickness has played a vital role in the evolution of the various bivalve clades, by determining the pathways open to them. Periostracal thickness has been important in the evolution of specific specialized life habits and of different styles of ornamentation. A similar study may also prove enlightening in investigating the evolution of other invertebrate groups. As noted in the introduction, the basic shell secretion mechanism used by the Bivalvia is used by all Recent members of shell-bearing molluscan classes, with the exception of the Polyplacophora. The possession of a periostracum and this mechanism can be regarded as primitive 86 PALAEONTOLOGY, VOLUME 40 for the phylum, and it seems intuitively obvious that all ancient members of these groups would have had periostraca. In particular, a study of the variation of periostraca in Recent gastropods (which do seem to have a great range of thickness) and any link that this might have with styles of ornamentation and life habit might be particularly rewarding. Additionally, there is scope for consideration of the brachiopods. Members of this phylum also secrete their shells on to an outer organic sheet, also termed the periostracum (see Williams and Mackay 1979). The structure is analogous but not homologous to the molluscan periostracum, and there are key differences in the manner in which the mineralized shell material is applied to the periostracum, most notably that it is secreted directly by the cells in the outer mantle lobe rather than by an extrapallial fluid. However, many aspects, such as the flexibility of the periostracal sheet and its implications for micro- ornament, must be similar to those found in molluscs and, therefore, the effects of variation of the periostracum are worthy of investigation. Acknowledgements. The equipment and travel costs for the visit to Panama for this project were met by a grant from the NERC (G R9 /II 09). Other living bivalves were collected during the tenure of a Visiting Fellowship to the Australian Museum and visits to the Swire Institute of Marine Science (Hong Kong) and Dunstaffnage Marine Laboratory (Oban, Scotland). Preserved material was collected from The Natural History Museum (London). I am indebted to all these institutions and to the many people in each, in particular Arturo Dominici in Panama, who expended much time and effort in helping me collect the variety of bivalves necessary for this study. The work was performed whilst a Research Fellow at Gonville and Caius College, Cambridge and a Royal Society University Research Fellow. J. Alphey assisted greatly with the understanding of the mechanics of sheet folding and Camp Stokell redrafted Text-figure 1 This is Cambridge Earth Sciences Publication 4641. REFERENCES aller, r. c. 1974. Prefabrication of shell ornamentation in the bivalve Laternula. Lethaia , 7, 43-56. araujo, r., ramos, M. a. and bedoya, J. 1994. Microtubules in the shell of the invasive bivalve Corbula fluminea (Muller, 1774) (Bivalvia: Heterodonta). Journal of Molluscan Studies, 60, 405 — 4 1 3 . baranthi, m. s. and ramalingham, k. 1989. Sclerotization of the periostracum of the marine bivalve Perna viridis (Linnaeus). Proceedings of the Indian Academy of Science, 98, 243-248. beedham, G. e. and owen, G. 1965. The mantle and shell of Solemya parkinsonia (Protobranchia: Bivalvia). Proceedings of the Zoological Society of London, 145, 405-430. bevelander. G. and nakahara, H. 1967. An electron microscopic study of the formation of the periostracum of Macrocallista maculata. Calcified Tissue Research , 1, 55-67. bottjer, d. j. 1981. Periostracum of the gastropod Fusitriton oregonensis : natural inhibitor of boring and encrusting organisms. Bulletin of Marine Sciences, 31, 916-921. — and carter, j. G. 1980. Functional and phylogenetic significance of projecting periostracal structures in the Bivalvia (Mollusca). Journal of Paleontology, 54, 200-216. bubel, a. 1976. An electron microscope study of the formation of periostracum in the freshwater bivalve Anodonta cygnea. Journal of Zoology, London, 180, 211-232. carriker, m. r., palmer, r. e. and prezant, r. s. 1980. Functional ultramorphology of the dissoconch valves of the oyster Crassostrea virginica. Proceedings of the National Shellfisheries ’ Association, 70, 139-183. carter, j. G. 1978. Ecology and evolution of the Gastrochaenacea (Mollusca, Bivalvia) with notes on the evolution of the endolithic habitat. Bulletin of the Peabody Museum of Natural History , 41, 1-92. — 1990. Skeletal biomineralization : patterns, processes and evolutionary trends. Van Nostrand and Reinhold, New York, 832 pp. — and aller, r. c. 1975. Calcification of the bivalve periostracum. Lethaia, 8, 315-320. carter, r. m. 1967. The shell ornament of Hysteroconcha and Hecuba (Bivalvia): a test case for inferential functional morphology. The Veliger, 10, 59-71. checa, a. 1995. A model for the morphogenesis of ribs in ammonites inferred from associated microsculptures. Palaeontology, 37, 863-888. Clark, G. r. 1976. Shell growth in the marine environment: approaches to the problem of marginal calcification. American Zoologist, 16, 617-626. crampton, j. s. 1990. A new species of Late Cretaceous wood boring bivalve from New Zealand. Palaeontology, 33, 981-992. HARPER: MOLLUSCAN PERIOSTRACUM 87 davies, d. j., powell, E. n. and stanton, R. J. 1989. Relative rates of shell dissolution and net sediment accumulation - a commentary : can shell beds form by gradual accumulation of biogenic debris on the sea floor? Lethaia, 22, 207-212. dunachie, j. F. 1963. The periostracum of Mytilus edulis. Transactions of the Royal Society, Edinburgh, 65, 383—41 1 . gray, j. e. 1825. Conchological observations, being an attempt to fix the study of conchology on a Arm basis. Zoological Journal (London), 1, 204-223. gregoire, l. 1974. On the organic and mineral components of the shells of the Aetheriidae. Revue de Zoologie Africaine, 88, 847-896. harper, e. M. 1991. The evolution of the cemented habit in the bivalved molluscs. Unpublished Ph.D. Thesis, The Open University. — 1992. Post-larval cementation in the Ostreidae and its implications for other cementing Bivalvia. Journal of Molluscan Studies, 58, 37-48. and skelton, p. w. 1993a. The Mesozoic Marine Revolution and epifaunal bivalves. Scripta Geologica , Special Issue, 2, 127-153. — 19936. A defensive value of the thickened periostracum in the Mytiloidea. The Veliger, 36, 36-42. hook, J. E. and golubic, s. 1988. Mussel periostracum from deep-sea redox communities as a microbial habitat : the scalloping periostracum borer. Marine Ecology, 9, 347-364. Hudson, J. d. 1968. The microstructure of a Jurassic mytilid (Bivalvia). Palaeontology, 11, 163-182. kawaguti, s. and ikemototo, n. 1962. Electron microscopy of the mantle of a bivalve, Fabula nitidula. Biological Journal of Okayama University , 8, 21-30. Kennedy, w. j., taylor, J. D. and hall, A. D. 1969. Environmental and biological controls on bivalve shell mineralogy. Biological Reviews, 44, 499-530. kniprath, E. 1979. The functional morphology of the embryonic shell-gland in the conchiferous molluscs. Malacologia, 18, 549-552. kri'z, J. 1972. Periostracum in Palaeozoic Bivalvia? Nature Physical Science, 237, 28-30. logan, a. 1974. Morphology and life habits of the Recent cementing bivalve Spondylus americanus Hermann from the Bermudan Platform. Bulletin of Marine Science , 24, 568-594. lutz, r. a., kennish, m. j. and tan, a. s. 1994. Shell microstructure of the vesicomyid clam, Calyptogena magnifica. Malacological Review, 27, 75-85. martill, d. and harper, l. 1990. An application of critical point drying to the comparison of modern and fossilized soft tissues of fishes. Palaeontology, 33, 423+428. moore, j. d. and trueman, r. e. 1971. Swimming of the scallop Chlamys opercularis. Journal of Experimental Marine Biology and Ecology, 6, 179-185. moore, r. c. 1969. (ed.). Treatise on invertebrate palaeontology. Part N. Mollusca 6. Bivalvia. The Geological Society of America and the University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 952 pp. morton, b. 1974. Some aspects of the biology and functional morphology of Cleidothaerus maorianus Finlay (Bivalvia : Anomalodesmata : Pandoracea). Proceedings of the Malacological Society of London , 41, 201-222. — 1979a. The biology, functional morphology and taxonomic status of Gaimardia ( Neogaimardia ) finlayi (Bivalvia: Gaimardiidae). Journal of Zoology, London, 188, 123-142. — 19796. Some aspects of the biology and functional morphology of Trapezium ( Neotrapezium ) sublaevigatum (Lamarck) (Bivalvia: Arcticacea). Pacific Science, 33, 177-194. — 1980a. Some aspects of the biology and functional morphology (including the presence of a ligamental lithodesma) of Montacutona compacta and M. olivacea (Bivalvia: Leptonacea) associated with coelenterates in Hong Kong. Journal of Zoology , 192, 431-455. — 19806. Some aspects of the biology and functional morphology of Coralliophaga ( Coralliophaga ) coralliophaga (Gmelin 1791) (Bivalvia: Arcticacea): a coral associated nestler in Hong Kong. 311-330. In morton, b. (ed.). Proceedings of the First International Workshop on the Malacofauna of Hong Kong and Southern China, Hong Kong. Hong Kong University Press, Hong Kong, 345 pp. — 1987a. The mantle margin and radial mantle glands of Entodesma saxicola and E. inflata (Bivalvia: Anomalodesmata: Lyonsiidae). Journal of Molluscan Studies, 53, 139-151. — 19876. The functional morphology of Neotrigonia margaritacea (Bivalvia: Trigoniacea) with a discussion of phylogenetic affinities. Records of the Australian Museum , 39, 339-354. — 1990. Corals and their bivalve borers -the evolution of a symbiosis. 1^16. In morton, b. (ed.). The Bivalvia - Proceedings of a Memorial Symposium in Honour of Sir Charles Maurice Yonge , Edinburgh, 1986. Hong Kong University Press, 355 pp. — 1992. The Bivalvia: future directions for research. American Malacological Bulletin , 92, 107-116. PALAEONTOLOGY, VOLUME 40 1996. The evolutionary history of the Bivalvia. 337-356. In taylor, j. d. (ed.). Origin and evolutionary radiation of the Mollusca. Oxford University Press, Oxford, xiv + 392 pp. nicol, D. 1965. Ecological implications of living pelecypods with calcareous spines. Nautilus , 78, 109-116. 1983. Nestling pelecypods. Florida Scientist, 46, 57-59. ockelmann, k. w. 1983. Descriptions of mytilid species and definition on the Dacrydiinae (n. subfam.) (Mytilacea: Bivalvia). Ophelia, 22, 81-123. pojeta, j. 1978. Review of Ordovician pelecypods. Professional Paper of the United States Geological Survey, 695, 1-46. prezant, r. s 1981. The arenophilic radial mantle glands of the Lyonsiidae (Bivalvia: Anomalodesmata) with notes on lyonsiid evolution. Malacologia, 20, 267-289. saleuddin, a. s. m. 1974. An electron microscope study of the formation and structure of the periostracum in Astarte (Bivalvia). Canadian Journal of Zoology, 52, 1463-1470. 1980. Shell formation in molluscs with special reference to periostracum formation and shell regeneration. 47-81. In spoel, s. van der, bruggen, a. van and lever, j. (eds). Pathways in malacology. Scheltema and Holkema, Utrecht, 294 pp. and petit, h. p. 1983. The mode of formation and the structure of the periostracum. 199-234. In saleuddin, a. s. m. and wilbur, k. m. The Mollusca. Volume 4 : Physiology, Part I. Academic Press, New York, 523 pp. savazzi, E. 1985. Adaptive themes in cardiid bivalves. Neues Jahrbuch fur Geologie und Paldontologie, Abhandlungen, 170, 291-321. Skelton, p. w. 1985. Preadaptations and evolutionary innovations in rudist bivalves. Special Papers in Palaeontology , 33, 159-173. — crame, j. a., morris, n. j. and harper, e. m. 1990. Adaptive divergence and taxonomic radiation in post- Palaeozoic bivalves. 91-1 17. In taylor, p. d. and larwood, g. p. (eds). Major evolutionary radiations. The Systematics Association Special Volume 42. Clarendon Press, Oxford, 437 pp. Stanley, s. M. 1968. Post-Paleozoic adaptive radiation of infaunal bivalve molluscs - a consequence of mantle fusion and siphon formation. Journal of Paleontology , 42, 214-229. - 1970. Relation of shell form to life habits of the Bivalvia. Memoirs of the Geological Society of America, 125, 1-296. taylor, j. d. and Kennedy, w. j. 1969. The influence of the periostracum on the shell structure of bivalve molluscs. Calcified Tissue Research, 3, 274-283. — and hall, a. d. 1969. The shell structure and mineralogy of the Bivalvia. Introduction. Nuculacea-Trigoniacea. Bulletin of the British Museum ( Natural History), Zoology Series, Supplement , 3, 1-125. — 1973. The shell structure and mineralogy of the Bivalvia. II. Lucinacea-Clavagellacea, Conclusions. Bulletin of the British Museum ( Natural History), Zoology Series, 22, 225-294. tevesz, m. j. and carter, j. g. 1980. Environmental relationships of shell form and structure of unionacean bivalves. 295-322. In rhoads, d. c. and lutz, r. a. (eds). Skeletal growth of aquatic organisms. Plenum Press, New York and London, 750 pp. thomas, R. d. k. 1978. Limits to opportunism in the evolution of the Arcoida (Bivalvia). Philosophical Transactions of the Royal Society of London, Series B, 284, 335-344. vermeij, G. j. 1987. Evolution and escalation. An ecological history of life. Princeton University Press, Princeton, New Jersey, 537 pp. waite, j. h. and anderson, s. o. 1980. 3,4-dihydroxyphenylalanine (DOPA) and sclerotization of periostracum in Mytilus edulis L. Biological Bulletin, 158, 164-173. waller, T. r. 1972. The functional significance of some shell microstructures in the Pectinacea (Mollusca: Bivalvia). Proceedings of the International Geological Congress, 24th Session, Montreal, Canada, Section 7, Palaeontology, 48-56. 1980. Scanning electron microscopy of shell and mantle in the order Arcoidea (Mollusca: Bivalvia). Smithsonian Contributions to Zoology, 313, 1-58. watabe, n. 1984. Shell. 448^485. In bereiter-hahn, j., mtoltsy, a. g. and Richards, k. s. (eds). Biology of the integument. 1, invertebrates. Springer-Verlag, New York, 825 pp. williams, a. and mackay, s. 1979. Differentiation of the brachiopod periostracum. Palaeontology, 22, 721-736. wright, m. w. and Francis, l. 1984. Predator deterrence by flexible extensions of the horse mussel Modiolus modiolus. The Veliger, 27, 140-142. yonge, c. m. 1977. Form and evolution in the Anomiacea (Mollusca: Bivalvia) - Pododesmus, Anomia, HARPER: MOLLUSCAN PERIOSTRACUM 89 Enigmonia (Anomiidae) and Placunomia, Placuna (Placunidae). Philosophical Transactions of the Royal Society of London , 276, 453-523. and Thompson, T. E. 1976. Living marine molluscs. Collins, London, 288 pp. ELIZABETH M. HARPER Department of Earth Sciences Typescript received 29 September 1995 University of Cambridge Revised typescript received 7 March 1996 Downing Street Cambridge CB2 3EQ, UK APPENDIX Average periostracal thicknesses and locality data for each species investigated herein. Where the data have been extracted from the literature, either as values given by other authors or determined from published micrographs, the reference is given. Taxon Average periostracal thickness (gm) Locality Protobranchs Acila divaricata 42 Arabian Coast Ennucula obliquaa 15 Malabar, N.S.W., Australia Leda minuta 3 Loch Goil Malletia obtusa 8 Norway Nucula nitida 10 Millport, Scotland Nucula nucleus 3 Millport, Scotland Nucula sulcata 25 Millport, Scotland Nucula turgida 10 Unknown Saturnia sulcata 10 Argentina Solemya australia 100 Taylor et al. (1969) Solemya borealis 100 Maine, USA Solemva occidentalis 29 Kenya Solemya parkinsoni 140 Beedham and Owen (1965) Solemva velum 10 Rhode Island, USA Yoldia eightsei 31 Signy Island, Antarctica Yoldia hyperbola 23 Greenland Yoldia thracinae 13 North-west Atlantic Yoldiella sabrina 5 Weddell Sea, Antarctica Arcoids Anadara erthraensis 57 Unknown Anadara ferrugianea 10 Gulf of Papua Anadara grand is 8 Panama Anadara trapeziana 100 Port Jackson, N.S.W., Australia Area tortuosa 17 Unknown Arcopsis solida 10 Naos, Panama Barbatia rostae 35 Naos, Panama Barbatia helbergi 8 Kenya Barbatia obliqua 200 Unknown Barbatia sp. 10 Panama Glycymeris flamea 2 Two Fold Bay, N.S.W., Australia Glycymeris glycymeris 11 North Sea Glycymeris holosericus 45 Jervis Bay, N.S.W., Australia Glycymeris maculata 24 Bottjer and Carter (1980) Limatula hodgsoni 1 Weddell Sea, Antarctica Limopsis enderbyensis 9 Weddell Sea, Antarctica 90 PALAEONTOLOGY, VOLUME 40 Average periostracal Taxon thickness (//m) Locality Limopsis marioensis 5 Modiolarca tulipa 14 Scapharca globosus 100 Scapharca inaequivalvis 54 Mytiloids Adipicola pelagica 15 Adula californianus 47 Adula falcata 19 Amygdalum beddomei 32 Aulacomya ater 81 Austromytilus rostratus 75 Bathymodiolus sp. 1 10 Bathymodiolus thermophilus 15 Botula cinnamonea 24 Botula siliqua 50 Brachidontes granulatus 15 Brachidontes niger 6 Brachidontes rostratus 35 Brack idon tes variabilis 15 Choromytilus chorus 400 Cibiticola lunata 5 Crenella decussata 5 Crenella discors 10 Crenella glandula 11 Fluviolamatus amarus 14 Geukensia demissa 92 Hormomya mutabilis 15 Ischadium recurvum 15 Limnoperna sinensis 28 Lioberus castaneus 22 Lioberus salvadoricus 8 Lithophaga sp. 45 Lithophaga aristata 8 Lithophaga cumingiana 29 Lithophaga erthraensis 27 Lithophaga kuehneti 11 Lithophaga kuentienti 5 Lithophaga lima 14 Lithophaga nasuata 25 Lithophaga nigra 21 Lithophaga teres 16 Modiolus indet. 10 Modiola striatula 53 Modiolus americanus 7 Modiolus areolatus 13 Modiolus barb at us 25 Modiolus capax 80 Modiolus elongatus 60 Modiolus metcalfei 30 Modiolus modiolus 50 Modiolus pseudotulipus 35 Musculista senhausia 34 Weddell Sea, Antarctica Millport, Scotland Unknown Unknown Shetlanc Washington (Carter 1990) New Zealand Cape Banks, Sydney, N.S.W., Australia Peru (Carter 1990) Two Fold Bay, N.S.W., Australia Hook and Golubic (1988) East Pacific (Carter 1990) Kenya Marlborougn Sound, New Zealand Bottjer and Carter (1980) Lagos, Nigeria Victoria, Australia Tai Tam, Hong Kong Chile Cape Ginger Carter (1990) Unknown Massachusetts, USA Woolgoolga, N.S.W., Australia North America Cape d'Aguilar, Hong Kong West Florida (Carter 1990) China Bottjer and Carter (1980) Panama Telegraph Bay, Hong Kong Naos, Panama Masirah Island, Oman Red Sea Oman Oman Oman Aldabara Carter (1990) Addu Aldabra Galeta, Panama Siput Fedo, Malaysia West Atlantic (Carter 1990) Two Fold Bay, N.S.W., Australia Unknown Naos, Panama Moreton Bay, Queensland, Australia Wu Kwai Sha, Hong Kong Firth of Lorn, Scotland Coco del Mar, Panama Tai Tam Bay, Hong Kong HARPER: MOLLUSCAN PERIOSTRACUM 9] Taxon Average periostracal thickness (/mi) Locality Musculus cumigianus 20 Narrabean Beach, N.S.W., Australia Musculus laevigatus 428 Hokkaido, Japan Musculus marmatus 10 Unknown Musculus nanus 50 Two Fold Bay, N.S.W., Australia Mytella guayensis 15 Panama (Carter 1990) Mytilospsis domingensis 4 Grand Cayman Mvtilus californianus 294 La Jolla, California Mytilus edulis 35 Two Fold Bay, N.S.W., Australia Parapholas californica 20 Carter (1990) Perna canaliculus 152 New Zealand Perna palliopunctatus 75 Bottjer and Carter (1980) Perna perna 160 Zululand, S. Africa Perna pictus 43 Algiers Perna viridis 45 Wu Kwai Sha, Hong Kong Septifer bilocularis 36 Wu Kwai Sha, Hong Kong Septifer virgatus 60 Cape d’Aguilar, Hong Kong Stavelia horrida 90 Keppel Bay, Queensland, Australia Stavelia torta 75 Unknown Trichomva hirsutus 100 Unknown Trisodos semi tort a 5 Unknown Unknown modiolid 37 San Felipe Market, Panama Xenostrobus securus 61 Port Stephen Pteriomorphs Adamussium colbeci < 1 Weddell Sea, Antarctica Aequipecten gibbosus < 1 Unknown Alectryonella crenulifera < 1 Hong Kong Alectryone/la haliotoidea < 1 Hoi Sing Wan, Hong Kong Amussium ballotti < 1 Broken Bay, N.S.W., Australia Amussium caudacum < 1 Zanzibar Amussium papiraceum < 1 Caribbean, Panama Amussium pleuronectes < 1 Unknown Anomia archaeus 8 Seychelles Anomia descripta < 1 Two Fold Bay, N.S.W., Australia Anomia ephippium 10 Galway, Eire Atrina inflat a < 1 Sai Kung Market, Hong Kong Atrina maura < 1 Panama Atrina pectinata < 1 Cape d'Aguilar, Hong Kong Atrina vexillum < 1 Zanzibar Chlamys asperrimus < 1 Victoria, Australia Chlamys opercularis < 1 Dunstaffnage Bay, Scotland Chlamys pusio < 1 Galway Bay, Eire Chlamys senatoria < 1 Seychelles Chlamys varia < 1 Galway Bay, Eire Crassostrea angulata < 1 Courseilles, France Crassostrea gigas < 1 Ministry of Agriculture and Fisheries Crassostrea virginica < 1 Carriker et al. (1980) Decatopecten plica L. < 1 Unknown Dimya corrugata < 1 Australia Electroma alacorvi < 1 Paula Salu, Singapore Enigmonia aenigmatica < 1 Singapore Equichlamys bifrons < 1 Coffin Bay, South Australia Foramelina exempla < 1 Sydney, N.S.W., Australia 92 PALAEONTOLOGY, VOLUME 40 Average periostracal Taxon thickness (//m) Locality Hinnites giganteus < 1 Hyotissa hyotissa < 1 Hyotissa latissimus < 1 Hyotissa numissima < 1 Hyotissa sinensis < 1 Isognomon bicolor < 1 Isognomon dentifer < 1 Isognomon ephippium < 1 Isognomon janus < 1 Isognomon legumen < 1 Isognomon perna < 1 Isognomon recognitus < 1 Lima colrata < 1 Lima excavata < 1 Lima fragilis < 1 Lima hians < 1 Lima lima < 1 Lima scabra < 1 Lyropecten antillarum < 1 Malleus regula < 1 Melina samoensis < 1 Mimachlamys gloriosa < 1 Monia squama 10 Neopycnodonte cochlear < 1 Neopycnodonte hyotissa < 1 Ostrea angasi < 1 Ostrea conchophila < 1 Ostrea edulis < 1 Ostrea irridescens < 1 Ostrea virescens < 1 Pecten diegensis < 1 Pecten fumatus < 1 Pecten jacobeus < 1 Pecten maximum < 1 Pedum spondyloideum < 1 Pinctada margaritacea < 1 Pinctada martensii < 1 Pinctada radiata < 1 Pinna bicolor < 1 Pinna carnea < 1 Pinna deltoides < 1 Pinna menkei < 1 Pinna rudis < 1 Pinna saccata < 1 Placunomia foliata 10 Plicatula plicata < 1 Plicatula imbricata < 1 Pseudamussium septemradiata < 1 Pteria brevialata < 1 Pteria atlantica < 1 Pteria chinensis < 1 Pteria penguin < 1 Saccostrea sp. < 1 Canadian Shellfisheries Addu Atoll, Maldives Maldives Oman Hong Kong Unknown Oman Tai Tam Harbour, Hong Kong Naos, Panama Cape d’Aguilar, Hong Kong Aldabara, Maldives Naos, Panama New Zealand Norway Hervey Bay, Queensland, Australia Millport, Scotland Grand Cayman Belize Grand Cayman Oman Ellice Island, Pacific Moreton Bay, Queensland, Australia Galway, Eire Atlantic Ocean Fairfax Island, Queensland, Australia Two Fold Bay, N.S.W., Australia Naos, Panama Galway Bay, Eire Naos, Panama Manly, N.S.W., Australia Clark (1974) Long Beach, Sydney, N.S.W., Australia Atlantic Ocean Millport, Scotland Shimoni, Kenya Aldabra Atoll Cape d'Aguilar, Hong Kong Addu Atoll, Maldives Zanzibar Grand Cayman Townsville, Queensland, Australia Vaucluse Bay Unknown New Caledonia Mediterranean Sea Tolo Channel, Hong Kong Phuket, Thailand Millport, Scotland Hong Kong Ghana Maldives Hoi Sing Wan, Hong Kong Galetos, Panama HARPER: MOLLUSCAN PERIOSTRACUM 93 Average periostracal Taxon thickness (//m) Locality Saccostrea commercialis < 1 Saccostrea cucullata < 1 Scaeochlamys livida < 1 Semipallium tigris < 1 Spondylus americanus < 1 Spondylus ducalis < 1 Spondylus marisrubri < 1 Spondylus sp. < 1 Stabilima strangei < 1 Streptopinna saccata < 1 Slriostrea margaritacea < 1 Vulsella vulsella < 1 Lucinoids Ambuscintilla praemium 2 Anodontia e den tula 10 Codakia tigerina 3 Corbis fimbriata 5 Ctena diver gans 5 Diplodonta codakia 8 Diplodonta diplodonta 10 Diplodonta globulosa 5 Diplodonta lateralis 5 Diplodonta te/linoides 6 Galeomma sp. 1 Kellia adamsi 5 Lasaea australis 3 Loripes clausus 2 Loripes lucinalis 3 Lucina pennsylvanica 5 Lucina rugifera 2 Montacutona compacta 5 Montacutona olivacea 5 Myrtea botanica 10 Myrtea spinifera 1 Parathysira resupina 8 Parvilucina fieldingi 3 Phacoides borealis 5 Unknown erycyinid 6 Heteroconchs Abra alba 30 Abra milaschewichi 2 Acanthocardia echinata 1 Americardia media 2 Angulus tenuis 1 Angulus tenuis 2 Anomalocardia squamosa 1 Arctica islandica 70 Asaphis deflorata 5 Astarte borealis 83 Astarte compressa 5 Lizard Island, Queensland, Australia Tai Tam Bay, Hong Kong Unknown Mombasa Tropical Shellfish Suppliers Tolo Channel, Hong Kong Oman Hong Kong Port Jackson, N.S.W., Australia Addu Atoll, Maldives Sri Lanka Cook Island, N.S.W., Australia Two Fold Bay, N.S.W., Australia Kenya Kenya New Caledonia Kenya Mirs Bay, Hong Kong Ghana North Heads, Sydney, N.S.W., Australia Kenya Naos, Panama Unknown Unknown Two Fold Bay, N.S.W., Australia Watanua, Kenya Isle of Wight, England Grand Cayman Shell Harbour, N.S.W., Australia Morton (1980a) Morton (1980a) Malabar, N.S.W., Australia Unknown Cronulla, N.S.W., Australia Oman Outer Hebrides, Scotland Unknown Millport, Scotland Black Sea, Rumania Millport, Scotland Unknown Oban, Scotland Millport, Scotland Hoi Sing Wan, Hong Kong Millport, Scotland Addu Atoll, Maldives Komandor Island, N. Pacific Arctic 94 PALAEONTOLOGY, VOLUME 40 Taxon Astarte elliptica Astarte sulcata Astarte willeti Atactodea glabrata At act ode a striata Bassina multilamellata Calyptogena magnifica Cardita affinis Cardita astartoides Cardita laticosta Cardita variegata Cerastoderma edule Cerastoderma glaucum Cerastoderma lamarkii Chama aspersa Chama buddiana Chama fibula Chama imbricata Chama jukesii Chama lazarus Chama reflexa Chama solida Chamalea gallica Chione mariae Chione subrugosa Circumphalus cassina Claudioconcha japonica Clausinella fasciata Clementia crassiplica Clinocardium nutalli Caecella chinensis Coelomactra antiqua Congeria cochleata Corbula crassa Corbula gibba Corbula smithiana Corbula sp. Crassatella florida Cutellus lacteus Cutellus hanleyi Cyclocardia australoidea Cyclocardia borealis Donacilla carnea Donax cueatus Donax faba Donax obesus Donax panamensis Donax scalpellum Donax variabilis Donax vittatus Dosinia dunkeri Dosinia exoleta Average periostracal thickness (/mi) Locality 26 29 10 1 1 1 60 1 10 2 2 2 1 1 1 1 1 1 1 1 1 1 20 1 1 3 1 3 1 2 4 14 7 89 100 100 50 2 20 16 2 2 17 8 10 2 2 10 3 25 10 10 Dunstaffnage Bay, Scotland Millport, Scotland Gulf of Alaska Addu Atoll, Maldives Addu, Maldives Hoi Sing Wan, Hong Kong Lutz et al. (1994) San Felipe Market, Panama Kengelan Islands Naos, Panama Tolo Channel, Hong Kong Dunstaffnage Bay, Scotland Sussex, England Wells, Norfolk, England Aldabra Atoll Naos, Panama Moreton Bay, Queensland, Australia Aldabara Atoll Great Barrier Reef, Australia Philippines Cape d’Aguilar, Hong Kong Naos, Panama Millport Panama Panama Millport, Scotland Tai Tam Bay, Hong Kong Millport, Scotland Broome, Western Australia Unknown Hong Kong Sai Kung Market, Hong Kong North Sea Tolo Channel, Hong Kong Millport, Scotland Two Fold Bay, N.S.W., Australia Panama, Naos Florida Keys Thailand Unknown Weddell Sea, Antarctica Bottjer and Carter (1980) Turkey Thailand Wu Kwai Sha, Hong Kong Panama Naos, Panama Oman Florida Oban San Felipe Market, Panama Millport, Scotland HARPER: MOLLUSCAN PERIOSTRACUM 95 Average periostracal Taxon thickness (//m) Locality Ensis directus 60 Ensis ensis 30 Ensis siliqua 50 Etheria elliptica 14 Eucrassaella kingicola 40 Eucrassatella cummingii 23 Eucrassatella cummingii 30 Fabula nitida 4 Florimetis cognatus 2 Gafrarium divaricatum 28 Gafrarium tumidum 5 Gaimardia finlayi 5 Gaimardia trapeziana 2 Gari fervensis 5 Glauconometta plankta 8 G/ossus humanus 100 Glossus vulgaris 15 Iphigena brasilliana 30 Kellia suborbicularis 10 Laevicardium crassum 70 Lutraria angustior 8 Macoma balthica 5 Macoma grandis 2 Macoma tent a 5 Macrocallista maculata 20 Mactra corallina 3 Mactra fonescana 10 Mactra laevicardia 8 Mactra mera 1 Meretrix costa 6 Meropesta nicobarica 10 Mulinia pallida 1 Nemocardia bechei 10 Neotrigonia bednalli 4 Neotrigonia gemma 3 Neotrigonia margaritacea 8 Neotrigonia sp. 12 Notocallista diemensis 3 Notospisula parva 1 Nutallia ezonis 10 Orbiculana orbiculata 33 Periglypta multicostata 2 Periglypta retisulcata 1 Petricola lucasana 2 Pharaonella perna 20 Pharella javanica 15 Pharella acutidens 4 Pharella japonica 18 Pharella jouanettia 10 Phaxas cutellus 5 Pitar tortuosa 1 Placamen molimen 2 Plagiocardia setosa 1 Narangasett Bay, LISA Dunstaffnage Bay, Scotland Millport, Scotland West Africa South West Australia Hervey Bay, Queensland, Australia N.S.W., Australia Kawaguti and Ikemoto (1962) San Felipe Market, Panama Cape d'Aguilar, Flong Kong Hoi Sing Wan, Hong Kong Morton (1979a) Discovery Expedition Millport, Scotland Two Fold Bay, N.S.W., Australia Lyn of Lorn, Scotland Unknown Florida, USA Friday Harbour, Washington, USA Plymouth, England Northern Atlantic Ocean Wexford, Eire Panama Trinidad Bevelander and Nakahara (1967) Scotland San Felipe Market, Panama Cockle Cove, Patagonia Tolo Channel, Hong Kong Unknown Oman Bicque, Panama Seychelles St Francis Island, Southern Australia Malabar, Sydney, N.S.W., Australia Morton (19876) Taylor et al. (1969) Long Bay, N.S.W., Australia Gosford, N.S.W., Australia Hokkaido, Japan Penang, Malaysia Panama Yeppoon, Queensland, Australia Naos, Panama Funzi, Kenya Unknown Tsu Sha Tsui Market, Hong Kong Sungei, Malaysia Sungei, Malaysia Kenya Panama Long Bay, N.S.W., Australia Hervey Beach, Queensland, Australia 96 PALAEONTOLOGY, VOLUME 40 Average periostracal Taxon thickness (//m) Locality Plebidonax deltoidea 5 Polymesoda inflata 23 Prototharca grata 1 Prototharca megintyi 1 Psammotellina semmelinki 15 Psamorbid sp. 12 Pseudopythina subsinuata 14 Quadrans gargadia 10 Quidnipagus palatau 3 Quidnipagus palatum 2 Raeta plicatella 2 Sanguinolatia donacioides 8 Schizothaerus sp. 70 Scintilla sp. 2 Scintilla striata 1 Scrobicularia plana 2 Sinovacula sp. 6 Solecurtus chamasolen 5 Solecurtus divaricatus 10 Solen cylindraceus 8 Solen marginatus 10 Solen regularis 100 Solen sicarius 62 Solen sp. 10 Solen vitreus 4 Soletellina vitatacea 4 Sphenia binghami 5 Spisula calcar 5 Spisu/a elliptica 30 Spisula solida 10 Spisula subtruncata 2 Strigilla eutronia 5 Tagelus dombei 63 Tagelus politus 16 Tagelus sp. 50 Tapes philippinarum 5 Tellina deltoidalis 5 Tellina donacina 1 Tellina donacina 2 Tellina fabula 2 Tellina tenuis 4 Theora fragilis 10 Tivela compressa 10 Trachycardium maculosum I Trachycardium sp. 1 Transenella puella 2 Trapezium sublaevigatum 110 Tridacna croce a 2 Tridacna maxima 1 Venericardia amabilis 2 Venerupis pullastra 5 Venerupis senegalensis 25 Venus ovata 1 Woolgoolga, N.S.W., Australia San Francisco, Panama Panama Panama Thailand Clyde River, N.S.W., Australia Hong Kong Aldabra Kenya Maldives Sanibal Island Lake Irrawarra, N.S.W., Australia British Columbia Tolo Channel, Hong Kong Oman Millport, Scotland China Unknown Carter (1990) Port Elizabeth Unknown Thailand Tofino, British Columbia Tai Tam, Hong Kong Unknown Kokan, India Oban, Scotland Unknown Millport, Scotland Millport, Scotland Millport, Scotland Ingham, Queensland, Australia Panama San Felipe Market, Panama San Felipe Market, Panama Hoi Sing Wan, Hong Kong Two Fold Bay, N.S.W., Australia Unknown Atlantic Ocean Atlantic Ocean Scotland Queensland, Australia Kenya Northern Oman Addu Atoll, Maldives Naos, Panama Morton (19796) Lizard Island, Queensland, Australia Cook Islands Long Bay, N.S.W., Australia Loch Torridon, Scotland Millport, Scotland Addu Atoll, Maldives HARPER: MOLLUSCAN PERIOSTRACUM 97 Average periostracal Taxon thickness (/mi) Locality Vepricardium asiaticum 1 Unknown Vepricardium sinensis 1 Unknown Anomalodesmatans Aspidopholas objecta 23 Lau Fau Shan, Hong Kong Cleidothaerus albidus 7 N.S.W., Australia Cleidothaerus maorianus 10 Morton (1974) Cochlodesma praetenua 3 Northumberland, England Coralliophaga coralliopbaga 15 Morton ( 1 980A) Corbula porcina 3 Montevideo, Uruguay Cuspidaria latesulcata 10 Cape Banks, Sydney, N.S.W., Australia Cuspidaria rostratus 10 Norway Cuspidaria tenella 5 South Orkney Cyrtodaria siliqua 33 N. Europe Ectorisma granulata 2 Malabar, Sydney, N.S.W., Australia Fulvia mutica 1 Tai Tam Bay, Hong Kong Gastrochaena cuneiformis 8 Kenya Gastrochaena dubia 18 Eire Gastrochaena hians 10 Carter (1978) Gastrochaena mytiloidea 10 Kenya Gastrochaena ovata 8 Carter (1978) Hiatella arctica 40 Millport, Scotland Hiatella australis 2 Two Fold Bay, N.S.W., Australia Hiatella orientalis 11 Hoi Sing Wan, Hong Kong Hiatella striata 8 Fowey, Cornwall, England Jouannetia cumingi 2 Phuket, Thailand Laternula elliptica 10 Unknown Laternula flexuosa 3 Aller (1974) Laternula gracilis 5 Western Point, Victoria, Australia Laternula moratina 2 Lake Macquarie, N.S.W., Australia Lyonsia norwegica 80 Northumberland, England Lyonsia californica 5 San Juan Island Lyonsia hyalina 3 Naragansett Bay, USA Martesia striata 10 Cape d'Aguilar, Hong Kong Mya japonica 15 China Mya truncata 20 Dunstaffnage Bay, Scotland Myadora complexa 15 Sydney, N.S.W., Australia Myochama anomioides 14 N.S.W., Australia Myochama strangei 60 Port Jackson, N.S.W., Australia Offadesma angasi 3 New Zealand Parapholas quadrizanatta 8 Oman Pholas sp. 10 Naos, Panama Pholas dactylus 5 Margate, Kent, England Pholas parva 10 Devon, England Saxicava rugosa 20 Brighton, England Spengleria rostrata 15 Carter (1978) Sphenia fragilis 10 Naos, Panama Thracia beningi 5 Cook Inlet, Alaska Xylophaga dor sal i a 10 Unknown THE JURASSIC AMMONITE IMAGE DATABASE ‘AMMON’ by bo liang and Paul l. smith Abstract. "Ammon' is an interactive database that incorporates taxonomic, morphological, stratigraphical and locality information as well as digitized images. When necessary, a user can measure morphological features directly from an image using the mouse to mark points on which are based the measurement of distances, the calculation of ratios and derivation of logarithmic spiral parameters. Features such as ribs can be counted using the mouse, and counts extrapolated automatically to a standardized unit of measurement such as a half whorl. ‘Ammon’ is also provided with a module that can measure features automatically once the image has been processed using edge detection, line thinning and tracing algorithms. The characterization of whorl shapes is a difficult problem best addressed by the use of elliptic Fourier analysis which not only faithfully mimics whorl cross sections but can also be animated to show, for example, the transition from one shape to another during ontogenetic development. 'Ammon' is a useful aid in fossil identification both as a retriever of species based on specified morphological, stratigraphical, and/or geographical information and as a way of assembling image collages of specified families, genera or species. Several general issues are raised by advances in computer technology and the growth in use of "Ammon’ and its kind. Paramount are: the universality of database design; completeness of taxonomic, stratigraphical and geographical coverage; efficiency of data entry; and the question of accessibility. While palaeontologists are quick to acknowledge the imperfections of the fossil record, they point less readily to the almost overwhelming wealth of still accumulating data. The collective database is so large and complex that in the past it could only be dealt with by specialists operating within spheres of taxonomic, temporal and spatial competence. Evolving technology is softening and perhaps even breaking down the boundaries of these spheres, allowing us to undertake research of a style and content that was not previously possible. Witness the impact, so to speak, of taxonomic databases in the area of extinction studies (Raup and Sepkoski 1982; Sepkoski 1993). Certainly palaeontological databases are proving of considerable service to those engaged in geological mapping, museum management, and exploration for minerals and fuels, although educational applications have barely been considered (Huber 1990; Price 1984; Rich 1989). Databases built so far often deal with fairly coarse taxonomic and temporal units where morphology is usually not emphasized, images are rarely incorporated directly, and the computer is used as a sophisticated filing cabinet rather than as a means of generating new data. An important question in considering the application of computers to palaeontology is whether fossil identification can be done entirely by expert systems or whether, as we suggest below, a marriage of human and machine capabilities is the best approach. At a fundamental level, the description of morphology and the documentation of variation may be approached in two ways: (1) the use of geometric models to describe morphology quantitatively, an approach that is theoretically capable of producing a universal set (or morphospace) of all possible morphologies; (2) the use of specimens and their images to obtain quantitative and qualitative data, thereby determining which parts of the universal set are actually occupied. Computers and computerized image databases readily unite these approaches allowing us to chart the boundaries of occupied morphospace through evolutionary time, document the evenness of occupation and wonder about the functional or adaptational significance of unoccupied morphospace volumes (Gould 1991). Questions of function as well as genetic linkage are also raised [Palaeontology, Vol. 40, Part 1, 1997, pp. 99—1 1 2| © The Palaeontological Association 100 PALAEONTOLOGY, VOLUME 40 when covariation amongst morphological features is detected, an area where the sheer capacity of computers to manipulate and analyse data comes into its own. The prototype ‘Ammon’ was initiated to handle systematically data accruing from stratigraphical studies of Jurassic sedimentary basins in western North America (Smith 1986). It has been growing steadily in content and scope, and there is now an independent sister database called Goniat dealing with Palaeozoic ammonoids (Kullmann et al. 1993; Korn et al. 1994). The purpose of the present paper is to document recent changes to 'Ammon’ that demonstrate: (1) the incorporation of digitized images into a database; (2) an interactive module that helps a user measure morphological parameters directly from an image; (3) an automatic module where the computer measures and derives morphological parameters independently using edge detection techniques; (4) the potential of elliptic Fourier analysis for characterizing whorl shapes and ontogenetic change; (5) the use of an interactive database as an aid to fossil identification. Copies of the new computer programs and interfaces (modules) mentioned in this paper ( ImagEdit , Caliper , Imagie and Animator) have been deposited at the British Library, Boston Spa, Yorkshire, U.K., as Supplementary Publication No. Sup 00000. In a subsequent paper, ‘Ammon’ and its new modules will be used to reveal hitherto undetected patterns of covariation and morphological diversity in Lower Jurassic ammonites. AMMON’ The database ‘Ammon’ was originally designed for a mainframe computer using Taxir as the database management program (Smith 1986). Subsequently, it was transferred to a workstation (SparcStation) and converted to work under Oracle Relational Database Management System at which time several image related descriptors were added. The present greatly expanded version of ‘Ammon’ contains 7790 specimens representing 15 families, 179 genera and 1319 species. Each specimen has 102 descriptors covering taxonomy, quantitative morphology, qualitative mor- phology, stratigraphy, locality information and general comments (Table 1). These descriptors and their states have been described in detail by Smith (1986) except for the following new descriptors (Table 1): 13 SYNSPECIES is the valid species name as judged by the person operating the database; 21 WHMAX is the maximum measurable whorl height for incomplete specimens; 27 RIBWIDTH is the ratio of maximum rib width to whorl height; 31 FURCPOSO (the quantitative equivalent of FURCPOS) is the distance between the rib furcation point and the umbilical seam expressed as a ratio of the whorl height; 32 UNITUBPOSO (the quantitative equivalent of UNITUBPOS) is the distance between a tubercle and the umbilical seam in unituberculate ammonites expressed as a ratio of the whorl height; 35 VOLUTIONO (the quantitative equivalent of VOLUTION) is the ratio of whorl overlap to inner whorl height (ol/wh in Text-fig. 1); 36 AH is the ratio of whorl overlap to outer whorl height (ol/WH in Text-fig. 1); 82 AREA divides the world into a few broad geographical regions such as the Western Pacific, Western Tethyan, North America, South America, Northwest Europe so that rapid scans for biogeographical data can be made; 99 IMAGE and 100 SCALE give the relative path of image files in the file system and the scale of the illustration; 101 SPECFEATURE is added because some taxonomic groups have unusual features that are important for identification but not common enough to warrant a separate descriptor in the database, e.g. the mid-flank spiral groove in Hildoceras. As shown by Raup (1967), the basic geometry of planispiral ammonites can be described by just three parameters (Text-fig. 1): the whorl expansion rate (W), the umbilical ratio (U), and whorl compression (WWWH). The list of descriptors shown in Table 1, which includes the Raup parameters, is not rigidly fixed but expansion to accommodate new categories of data obviously requires all existing entries to be updated. In our unpublished study using ‘Ammon’ to examine covariation and recovery from the end-Triassic mass extinction, for example, we did not have to deal with heteromorph ammonites and consequently the Raup parameters were adequate for our needs. Some descriptors are derived from others and it is a matter of judgment as to whether they should be stored separately or simply calculated when needed thereby trading computer storage table 1. The 102 descriptors of the database 'Ammon' arranged in six categories as shown on the left. The descriptors have been defined (Smith 1986), with the exception of the 10 additions described in the text that either augment existing descriptors, quantify qualitative descriptors, or LIANG AND SMITH: AMMONITE IMAGE DATABASE 101 O z 5 F, Q j C/3 >H ^ ^ 2 n ® ^ o ffl S Z H Pa3c.>.-jo3;-Hm_i O Z u o -p o tx z u o 2; up 3 O* O' C/3 C/3 C/3 m C/3 3-3 >- Z ^ Z -1 Ul d up = O % O to 05 z < 3 3 >« Ph C/3 C/3 c/3 m ri PJ u Q < > °- > C/3 03 a: Pi J u UJ D pp 3 UP CQ H z H H pp D 3 D o < < 3 Ph H C/2 N Q -1 OO o SO H PJ o o 03 N o p cc ° Ph O 3|&H>3>Cpc/3 < £ pj d u < 0- ^ m Ph 7)0^a: M >0 OO -h Qj >H o pp UJ O C/2 S3 z o3 UJ 2 2 s z o oC w PQ u z o H U UP PJ £ < H oZ Qj o3 N •<5 > z hJ < X CP Z o o O 03 o u -P 2 oC Ph X UP Ph Uh Up 3 PJ 03 o O w C/2 C/2 UJ > Oh C/2 U UP 2 CP hJ u pj oc o no (N OO o so Cj MS bU cS £ 2 < UP CP < X C/3 “S^>. C/2 ^ p W X- jS999<<53hO 5“a;o3^a''_;OH<; &>T'I'>JEhl~DO«OLU/'(j Q?alu?pHa.c/3Uc/3tP Gs t/~ ) »-h ro Os i/~ ) o -HMMfOri^flTiSO'OhOOOOOOChO O C- c* < 03 j 03 > u up d pp i - Q S E fc; uj o3 < x 3 S V ca as < x z ^ >h UP 03 Q Z 20 ^SoQafesL«l§q o C/3 o z a- O ca h 3 b Dp5,<>13o,MOmK/3
  • , text-fig. 1. Ammonite cross section showing the basic geometric parameters (modified from Smith 1986). W = (WH/wh)2,I/r where r is the angular distance between WH and wh (in this case r = 2n because WH and wh are separated by one whorl); U = UD/D; WWWH = WW/WH. capacity for increased retrieval time. An important example is the length of the body chamber, a parameter that is necessary for understanding hydrostatics but difficult to use in systematics because the peristome is rarely preserved and often the body chamber itself is completely missing. When data are available, the angular length of the body chamber in radians (r) can be calculated from the expansion rate (W), the maximum shell diameter (DMAX), and the diameter of the phragmocone (DPHRAG), by rearranging the expression W = (DMAX/DPHRAG)2,t/r (Smith 1986). Ammonite illustrations are scanned into a Sun SparcStation using a MicroTek image scanner with 256 grey shades. Fossil plates in journal articles normally have multiple specimens figured on each plate. To speed up the digitization process, a whole plate is scanned into the computer and a program ( ImagEdit ) is used to extract individual specimens, clear irrelevant components such as figure numbers, and save the selected part of the image as a separate file. INTERACTIVE IMAGE MEASUREMENT MODULE When a specimen is entered into the database for the first time, morphological data are sometimes not available and have to be measured from images of the specimen. This is facilitated by a front- end user interface called Caliper built on top of the ‘Ammon’ database. Caliper is a module implemented in C, embedded SQL, Sunview and a graphics library called Pixrect. Its function is to derive morphological characters by combining the human ability to locate features visually with the computer’s ability to memorize and compute. With the lateral image of the ammonoid on the screen, the operator uses the mouse to position five control points: O on the coiling axis, P, and P2 on the inner coiling curve, and P3 and P4 on the outer coiling curve (Text-fig. 2). From these points, the shell expansion rate (W) and the umbilical ratio (U) are calculated and the inner and outer coiling curves simulated. By definition (Text-fig. 1) W = (OP3/OP4)2*/r, where OP3 and OP4 are separated by an angular distance of r radians which must be less than tc to make coiling direction unequivocal to the program. U = OP2/OP3. An equivalent measurement of W for the inner whorl is WI = OP2/OP4, where OP2 and OP4 are separated by one whorl. WI can be used to simulate the inner whorl spiral. The diameter of the shell and the whorl height at point P3 are: D = OP3 + OP3/V W, WH = P,P3. LIANG AND SMITH: AMMONITE IMAGE DATABASE 103 Retrieve by: Search : SPECNO = 43252a Images Returned: 3 Remained: 0 Whorls: Taxonomy => ] [w (out) =|1.9B fw (in) =|2.0B fUT[0.47 [MaxVH=| [Rib Width- ] 0 ■ 22 [ VW/WH 1 0.88 [ Furcpos= | [ Uni tubpos= [ PRHW- ] 15 [ SRHV= ] [THtf=] [WPT| 0.27 fcHF| |[T SUPERFAMILY:Eoderocerataceae | FAMILY: Li parocerati dae GENUS : Androgynoceras SPECIES : 1 ataecosta T AXAUTHYEAR : ( Sou . ) REFAUTHYEAR :Meister 1986 SYNONYMY : Androgynoceras SYNSPECIES : 1 ataecosta STAGE:P1 iensbachian SUBSTAGE :Upper/Loujer EURZONE :Davoei EURSUBZONE : Stokesi/Macul atum | SUBZONE :Maculatum-Stokes1 AREA:Western Tethyan COUNTRY: France PROVINCE :Causses basin SECTNAME :XVb L0CN0 :36 SPECNO :43252a IMAGE:/home/ images/meister /meister86pl7-l.ras D: 8.91 UD: 4.15 U: 0.47 EXP: 1.98 WH: 2.74 VHD: 0.31 WV0: 0.27 WVWH: 0.88 text-fig. 2. Data entry and interactive image measurement screen for ‘Ammon’ (colour suppressed). Descriptor states can be entered automatically by clicking on the relevant buttons in the top middle panel (those relevant to the left figured specimen are highlighted). The control points marked O, P and A-D, which are placed on the images by the user, are used to simulate whorl spirals, calculate quantitative parameters, and reconstruct specimen fragments. The top right panel shows all the data sorted for the left image. The left panel is the command window housing buttons for retrieving and manipulating data. The whorl overlap ratios with respect to the inner whorl and the outer whorl are (Text-fig. 2): VOLUTIONO = P2X/P,P2, AH = P2X/P2P3 To calculate the above parameters, all the user needs to do is to mark the five control points with the mouse then press one of the ’W(out) = ’, the ’W(in) =’ or the ‘ U = ’ command buttons in the top window. The program will superimpose the simulated inner and outer coiling curves on the image and show the parameter values in the text fields following the command buttons (Text-fig. 2). If a specimen is incomplete or damaged and the control points cannot be located with confidence, a ‘try and see’ approach can be used; the ‘Reset image’ button in the left window will dispose of unsuccessful attempts. Once the location of the coiling axis is settled, the primary rib density (PRHW) can be counted by clicking on a number of consecutive ribs with the middle mouse button. If N ribs are marked then PRHW = N*7r/r, 104 PALAEONTOLOGY, VOLUME 40 where r is the angular distance between the first and last ribs which must be less than n to avoid ambiguity about coiling direction yet large enough to justify extrapolation of the number of ribs to a half whorl (PRH W). This is a useful tool for using rib densities to characterize species and exploring primary rib density change during ontogeny. Secondary rib density (SRHW), if applicable, can be determined in the same way. Another group of parameters related to whorl shape can be measured from a cross section (Text- fig. 2, left). To measure whorl width (WW), the user clicks on points A and B, then presses the ‘Distance’ button in the left window (Text-fig. 2). To measure and calculate whorl compression (WWWH) and the fineness ratio (WWD), the user clicks on the four control points A through D and then presses the ‘Ratio’ button. For most specimens, cross section drawings are not available, so the above parameters have to be derived from ventral view images. Since the ventral image is subject to photographic distortion at both ends, the two control points for whorl width (A' and B' in Text-fig. 2) must be located in the centre of the view with line A'B' perpendicular to the plane of bilateral symmetry. Caliper then determines the shell diameter and whorl height at this location. When the control points C' and D' have also been located, the whorl width to shell diameter ratio and the whorl width to height ratio can be obtained by pressing the ‘WWD =’ and ‘WWWH =’ buttons in the top window. The text fields in Text-figure 2 show the same results as measured directly from the cross section, namely, WWD = 027 and WWWH = 0 88. The icon panel which is shown ‘pulled down’, in the centre of Text-figure 2 is designed to facilitate the input of qualitative morphological data into ‘Ammon’. It is evident from the images of the specimen that it has strong, straight and gently prorsiradiate ribs; a steep umbilical wall; a quadrate to rounded whorl section (the database ‘Ammon’ allows multiple states for any character of a specimen) and a plain venter. The user clicks on the corresponding icons which the program highlights. Pressing the ‘Save’ button in the left window stores all qualitative as well as quantitative morphological data without the user typing a single word or having to memorize any character states. AUTOMATIC IMAGE MEASUREMENT MODULE To explore the possibility of a fully automatic approach, a program called Imagic was written to measure shell expansion rate and rib density from lateral view images of specimens without human intervention. The program applies to specimens with simple ribs; other morphological features such as tuberculation and rib furcation are ignored to make the task tractable. The geometry and ornamentation of ammonites have a high degree of regularity and this is the basis of image interpretation since it allows the computer to ignore noisy elements of the image that reflect the state of preservation rather than shell morphology. Analysis first involves an image preparation stage after which the computer locates the coiling axis and measures several parameters. Image preprocessing Images are cleared of irrelevant components such as text labels using the program ImagEdit. To focus the subsequent analyses on the ammonite image, the average (X) and standard deviation (S) of the background brightness intensity are estimated from a three-pixel wide sample of the entire margin of the background. The program reassigns brightness intensity value (256 grey shades) to each pixel in the image as follows: Z = 0 when Z < X + S and X < 128 (black background), Z = 256 when Z > X — S and X > 128 (white background), Z = Z otherwise, where Z is the brightness intensity of the pixel. The major elements in an ammonite image are coiling curve and ribs; both are edges where brightness shows significant changes. Edge detection serves to simplify the analysis of images by drastically reducing the amount of data to be processed, while at LIANG AND SMITH: AMMONITE IMAGE DATABASE 105 A B text-fig. 3a. Edge detection output from the Canny operator used to locate the coiling axis from possible rib and coiling curve segments. The hypothetical coiling axis is marked O; a is the angle between the radius vector and the midpoint of a radially arranged line segment (rib) and /? is the angle between radius vector and the tangent to the spiral. The ammonite is a poorly preserved specimen of Dubariceras freboldi (from Smith et al. 1988). 3b. An example of automatic image measurement, where a matching coiling curve and rib pattern simulated by the computer is superimposed over the original specimen (Dubariceras freboldi, as used in a). The numbers represent calculated rib densities per half whorl at various stages of ontogeny. Scale bar represents 10 mm. the same time preserving useful structural information about object boundaries. The Canny operator (Canny 1986) was chosen for this purpose. The output of the Canny operator consists of separate pixels which are then linked into line segments according to 2-D spatial proximity and collinearity. Each line segment has three components: orientation, length in terms of number of edge points in the segment and co-ordinates of these points. Before the tracing of line segments, a thinning algorithm (Pavlidis 1982) is used to skeletonize edge segments potentially wider than one pixel into one-pixel wide segments. Text-figure 3a is an example of an ammonite image that has undergone edge detection, thinning and line tracing; the original image is shown in Text-figure 3b. Locating the coiling axis, coiling curve and ribs The most important feature of ammonite morphology is that ribs distribute around the coiling axis radially and coiling curves follow a logarithmic spiral starting from the coiling axis. The angle /?, which describes how rapidly the coiling curves move away from coiling axis, is made by the radius vector intersecting the tangent to the spiral. Lower Jurassic ammonites, for example, typically have ft values between 75° and 90° even when there are changes during ontogeny. A statistical parameter (SC) based on this observation is used to locate the coiling axis: m n SC = S C oiling Len[f\+ S RibLen[i\, 1 i= 1 where C oiling Len[i ] is the length of the ith line segment which has a /? value between 75° and 90° with respect to the hypothetical coiling axis O (Text-fig. 3a). RibLen[i\ is the length of the ith segment with a < 15° so the line segment is in a radial position to the hypothetical coiling axis, a 106 PALAEONTOLOGY, VOLUME 40 is calculated from the middle point of the segment (Text-fig. 3a). The SC value provides a measurement of support that the hypothetical coiling axis gets from all line segments in the image using the underlying geometric pattern. The coiling axis is nearly always located within a central region in the image which is half of the image size in terms of width and height. The program samples this region from left to right, top to bottom computing the SC values with an increment of 4 pixels which corresponds to 1 mm if the image scale is 1 and the monitor resolution is 100 dpi (dots per inch). The pixel which has the highest SC value is considered to be the location of the coiling axis. Distinguishing coiling curves and ribs is difficult because of noise, and line segments resulting from local brightness changes related to imperfections in specimen preservation (Text-fig. 3b). Again, the underlying geometric constraints of ammonite shell morphology are used to eliminate non-rib and non-coiling-curve elements using the following criteria: (1) empirically reasonable limits for the range of shell expansion rate can be set at 1-2-5, and possible coiling curve segments, and geometric relations among them, should meet this constraint; (2) coiling curves are nearly normal to radius vectors (/? > 75°); (3) ribs distribute radially around the coiling axis; (4) the length of each rib should be consistent with its neighbouring ribs; (5) the ‘white to black' rib edges should be sandwiched between two ‘black to white’ rib edges and vice versa ; (6) the width of each rib should be consistent with its neighbouring ribs. Generally speaking, statistical data on geometric features are more robust than individual measurements. The image is therefore divided into 12 equiangular sectors around the coiling axis. Rib and coiling curve segments are verified in each sector and then rechecked during the final assembling stage according to the above constraints. Deriving morphological parameters Once the coiling axis, coiling curves and ribs are determined, shell expansion rate (W), rib density (PRHW), volution (VOLUTIONO), whorl overlap (AH) and their changes during ontogeny can be computed easily. Nonlinear regression can be used to formulate rib forms. Text-figure 3b shows ribs and coiling curves simulated by the program superimposed on the original image; ribs were matched using the least squares technique. This program can be a useful tool for exploring ontogenetic changes. A user may determine the shell expansion rate between any two points by clicking on them with the mouse. Similarly, rib density between any two points may be determined by clicking on each rib. The program will display how many ribs there are between the two points and how many ribs there would be for half a whorl. By working from the inner to the outer whorl, the user can get a clear picture of how the shell expansion rate and rib density change during growth. In this example (Text-fig. 3b), the rib density gradually increases from 28 to 44 from the visible innermost whorl to the outer whorl, then starts to decrease to 34 on the last half whorl where, in this case, maturity is probably reached. ELLIPTIC LOURIER ANALYSIS OL WHORL SHAPE Whorl shape is an important aspect of ammonite shell morphology. A simple descriptor in the ‘Ammon’ database is the ratio of whorl width to whorl height (WWWH) which does not define the shape uniquely since a circle and a square would give the same ratio. Another descriptor WHORL_SHAPE uses descriptive terminology such as ‘quadrate’, ‘rounded’, ‘subrounded’, etc., but workers do not necessarily agree on the exact meaning of a given descriptive phrase such as ‘subrounded’, and it is difficult to compare variations across different taxonomic groups. These considerations create a need for a method like Fourier analysis which has been used in many palaeontological studies for the characterization of closed curves (Kaesler and Waters 1972; Anstey and Delmet 1973; Christopher and Waters 1974; Younker and Ehrlich 1977; Canfield and Anstey 1981; Foote 1989). One common Fourier method is polar Fourier analysis (Kaesler and Waters 1972) which we did not use because it is limited to simple curves without multiple re-entrants. We compared the more LIANG AND SMITH: AMMONITE IMAGE DATABASE 107 ONTOGENY Selected Whorl Shapes 1 3 4 4.5 5 7 REAL AMMONITE Ci Q 0 0 c r^\ FOURIER ANALYSIS Q Q 0 0 0 text-fig. 4. Ontogenetic change in whorl shape for Arnioceras ceratitoides, by whorl number. The upper sequence is taken from a real specimen (Blind 1963, fig. 25) whereas the matching sequence below is computer generated using elliptic Fourier analysis. From left to right, the number of harmonics used in the Fourier series are: 4, 5, 6, 8, 15, 23 (root mean square error < 0-01 ) reflecting the increase in morphological complexity during growth. sophisticated elliptic Fourier analysis (Kuhl and Giardina 1982) and perimeter-based Fourier analysis (Foote 1989) and found that elliptic Fourier analysis gave better results. There is also the advantage that there is no need to determine an artificial centroid for the shape. Furthermore, Elliptical Fourier descriptors are invariant with rotation, dilation and translation of the contour. Whorl shape drawings are scanned into the computer from published work. The previously mentioned thinning algorithm (Pavlidis 1982) is used to skeletonize lines potentially wider than one pixel and the line tracing program links separate pixels into a continuous outline ready for analysis. Text-figure 4 shows the ontogenetic variations of Arnioceras ceratitoides (from Blind 1963). The number of harmonics required to reduce root mean square error to 0 01 or less increases with ontogeny. This is consistent with the visual observation of the general increase in whorl shape complexity. The most significant change occurs from whorl 4-5 to 5 where the number of harmonics increases from 8 to 15, corresponding to the development of a keel. A program called Animator has been written to visualize and animate the detail of how one whorl shape can evolve into another. All whorl shapes in the sequence are produced at the same size and superimposed in sequence so that subtle changes in shape can be detected. In addition, Fourier analysis can detect and quantify whorl shape variations within and among taxa. It would be interesting to map the whorl shape distribution of naturally occurring ammonites with respect to their Fourier representations so that the relative density of occurrence can be evaluated. Current models of 3D simulation of ammonite shells assume a circular or elliptical whorl section (e.g. Raup 1967; Chamberlain 1981). Fourier representation of whorl sections make it possible to model real ammonites. 108 PALAEONTOLOGY, VOLUME 40 Family Psiloceratidae > ;Schlotheimidae > ! Arietitidae > i Echioceratidae > Oxynoticeratidae > !Cymbitidae > Eoderoceratidae > Coeloceratidae t> Phricodoceratidae > Polymorphitidae > Liparoceratidae P> Amaltheidae > j Dactylioceratidae l> Genus niiaocerdiiaae Catacoeloceras Phymatoceratidae Collina Graphoceratidae P Dacty/ioceras Sonniniidae P Nodicoeloceras Cardioceratidae P Peronoceras Erycitidae P Porpoceras Kosmoceratidae l> Preperonoceras Otoitidae P Prodactylioceras Sphaeroceratidae P Reynesoceras Stephanoceratidae P Reynesocoeloceras Oppeliidae P Zugodactylites Perisphinctidae > Reineckeiidae > Discophyliitidae > Phylloceratidae > Lytoceratidae > Family uncertain > text-fig. 5. Family menu (left) and genus submenu for the family Dactylioceratidae. AN AID TO IDENTIFICATION Identifying ammonites involves an evaluation of shell morphology and stratigraphical occurrence as well as an appreciation of geographical distribution. Once a list of potential candidate species has been formulated, it becomes necessary to compare numerous illustrations. This can be a difficult task if the list of candidates is long but, in addition, all species normally show variation which must also be evaluated before a species can be identified with confidence. Traditionally, intraspecific variation is assessed by examining the illustrations and descriptions of specimens in the synonymy of the species which may mean scouring a voluminous literature that is in several different languages and often reaches back well into the last century (Smith 1986). In order to utilize the literature effectively, the attributes of a large number of specimens have to memorized and evaluated in an objective manner. To a large extent, the image database ‘Ammon’ can overcome these problems. Given morphological information supplied by the user, a search of ‘Ammon’ is made and closely matching species displayed on the screen for comparison with the specimen in question. Other constraints such as the approximate stratigraphical range can be added to shorten the list of candidate species. Searches can also be restricted to a specified shell diameter range to circumvent the problem of ontogenetic change. ‘Ammon’ has an easy-to-use, menu-driven user interface which LIANG AND SMITH: AMMONITE IMAGE DATABASE 109 is written in Sunview and the embedded SQL query language of the Oracle Database Management System. Users can retrieve Jurassic ammonite data and images by using the 'Taxonomy’, 'Stratigraphy’, 'Geography’, ‘Morphology’, or ‘Reference’ buttons in the command window or any combination of these by using the ‘Combination’ button (Text-fig. 2, left). When the ‘Taxonomy’ button is pressed using the mouse, a pull-right menu appears which shows the available ammonoid families in the database. If, for example, the user selects the ammonite family Dactylioceratidae by highlighting it and releasing the mouse button, all specimens of the family in the database will be retrieved. Each item in the family menu has a pull-right menu producing a genus list for the family (Text-fig. 5). Suppose that the genus Dactylioceras is selected, then the screen shown in Text-figure 6 would be the query result. Image quality can be adjusted using the text-fig. 6. Retrieval results from 'Ammon' for the genus Dactylioceras. ‘Brightness’ and ‘Contrast’ sliders in the command window. The top-right window records the query text, gives the number of images retrieved and the number remaining for display. Images are displayed in order of decreasing size for efficient utilization of screen space; the user can go to next or previous screen of images by clicking the 'Next Page’ or 'Previous Page’ buttons in the command window. By clicking on an image, all information related to that specimen will be displayed in the middle-right text window (Text-fig. 6). A user interested in the species to which the specimen belongs and wanting to examine intraspecific variation can mark an image with the middle mouse button 110 PALAEONTOLOGY, VOLUME 40 and click on ‘Sped Sublist' which will display all the images for that species. Clicking on the ‘Main List’ button will return the display to the genus level. Exercises such as this enable anyone to familiarize themselves very quickly with a particular taxonomic group. The ‘Stratigraphy’ panel has a two-level menu. The top level is the Jurassic stage menu and each stage has a pull-right menu which shows the zones (north-west European scheme) in the stage. The user can scan the database by zone and get some idea of how ammonite morphology or ammonite communities changed through time. Exploring morphological and temporal changes of ammonite taxa in different parts of the world is possible by using ‘Geography’ as a query criterion, retrieving by geographical region or country. There are five geographical regions : Northwest Europe, Western Tethyan, Western Pacific, North America and South America, each of which has thousands of specimens in the database and could easily overwhelm the user. Geographical regions are most commonly used with other criteria to narrow a search. If the user has no idea as to which taxonomic group a specimen belongs, then the database can be searched using morphological characters such as the degree of involution, shell expansion rate, rib form and furcation pattern, tubercle pattern, whorl shape, ventral geometry, and so on. An example of pull-right menus for ribbing is given in Text-figure 7, which also demonstrates the access MorphoLogyj Rib l> " Tubercle * Whorl Shape 0 Venter > Expansion > Involution *> Rib Form l> :urcation > Form straight concave convex sinuous falcoid falcate biconcave projected text-fig. 7. Morphology menu (left) and pull-right submenus for ribbing showing selected descriptors and descriptor states in the qualitative morphology category of the database structure. routes to most of the descriptors in the qualitative morphology category of the database structure as indicated in Table 1. A user interested in a specific person’s work can query the database with the Reference button which will produce a list of authors arranged in alphabetical order. The ‘Taxonomy’, ‘Stratigraphy’, ‘Geography’, ‘Morphology’ and ‘Reference’ menu system offers easy access to the database but, at the same time, reduces flexibility because the user can only search the database in one area of inquiry at a time. For any serious application, a combination of the above search criteria is needed as provided by the ‘Combination’ panel in the command window (Text-fig. 6, left). DISCUSSION ‘Ammon’ can be useful in assisting with the identification of ammonites. The user only needs to know a few morphological descriptors and database operators. A new user of ‘Ammon’ may need to measure a few quantitative morphological parameters but after some practice they should be able to build up a sense of relationship between quantitative and qualitative parameters at which point identification becomes fun. The computer can even help by tabulating stored qualitative descriptors against their quantitative counterparts. Because of problems of poor preservation and the complex rules underlying taxonomy, it seems unlikely that computers will replace the human expert, but databases will alleviate the burden of memorizing many species and associated data allowing the LIANG AND SMITH: AMMONITE IMAGE DATABASE user to focus on more important questions. Continuing advances in computer hardware and software technologies have made it feasible to store and manipulate fossil images digitally with ease, and access speeds and storage capacities are progressively increasing while costs decrease. As this trend continues, image databases will proliferate and make a significant impact on the way that palaeontologists work, particularly if databases become networked through the ‘Information Superhighway’. It is not clear at this point whether these systems will simply continue to be created on an ad hoc basis and allowed to grow, compete and evolve, or whether palaeontological organizations will take a more proactive role and encourage efficient growth by avoiding duplication of effort, ensuring universal coverage, and guarding against restricted accessibility. We hope the latter because we cannot help but wonder whether the journal that you are holding in your hand is on the verge of extinction. Acknowledgements . We thank Dr R. Woodham of the Computer Science Department, UBC for providing the program of the Canny edge detector. This work was supported by an NSERC grant to Paul Smith and a UBC Graduate Fellowship to Bo Liang. REFERENCES anstey, r. l. and delmet, D. a. 1973. Fourier analysis of zooecial shapes in fossil tubular bryozoans. Bulletin of the Geological Society of America , 84, 1753-1764. blind, w. 1963. Die Ammoniten des Lias Alpha aus Schwaben, vom Fonsjoch und Breitenberg (Alpen) und ihre Entwicklung. Palaeontographica , Abteilung A , 121, 38-131. Canfield, d. j. and anstey, r. l. 1981. Harmonic analysis of Cephalopod suture patterns. Mathematical Geology , 13, 23-35. canny, j. 1986. A computational approach to edge detection. Institute of Electrical and Electronics Engineers Transactions on Pattern Analysis and Machine Intelligence , 8, 679-697 . chamberlain, j. a. 1981. Hydromechanical design of fossil cephalopods. 289-336. In house, m. r. and senior, j. r. (eds). The Ammonoidea. Systematics Association Special Volume 18. Academic Press, London, 593 pp. Christopher, r. a. and waters, J. a. 1974. Fourier series as a quantitative descriptor of miospore shape. Journal of Paleontology, 48, 697-709. dagis, a. a. 1968. [Toarcian Ammonites (Dactylioceratidae) from northern Siberia.] U.S.S.R. Academy of Sciences, Siberian Branch, Transactions of the Institute of Geology and Geophysics , 40. 1-108. [In Russian]. foote, m. 1989. Perimeter-based Fourier analysis: a new morphometric method applied to the trilobite cranidium. Journal of Paleontology, 63, 880-885. gould, s. J. 1991. The disparity of the Burgess Shale arthropod fauna and the limits of cladistic analysis: why we must strive to quantify morphospace. Paleobiology, 17, 411-423. huber, b. t. 1990. Digital-image processing will replace most hard-copy photography in palynology. Geotimes, 35, 43-44. kaesler, r. l. and waters, j. a. 1972. Fourier analysis of the ostracode margin. Bulletin of the Geological Society of America, 83, 1169-1178. korn, d., kullmann, j., kullmann, p. s. and petersen, m. s. 1994. Goniat, a computer-retrieval system for Paleozoic ammonoids. Journal of Paleontology, 68, 1257-1263. kuhl, f. p. and giardina, c. R. 1982. Elliptic Fourier features of a closed contour. Computer Graphics and Image Processing, 18, 236-258. kullmann, j., korn, d., kullmann, p. s. and petersen, M. s. 1993. The database system Goniat - a tool for research on systematics and evolution of Paleozoic ammonoids. Geobios, Memoire, 15, 239-245. meister, c. 1986. Les ammonites du Carixien des Causses (France). Memoires Suisses de Paleontologie, 109. 1-209. pavlidis, t. 1982. Algorithms for graphics and image processing. Computer Science Press, Rockville, Maryland, 416 pp. price, d. 1984. Computer-based storage and retrieval of palaeontological data at the Sedgwick Museum, Cambridge, England. Palaeontology, 27, 393-406. raup, d. m. 1967. Geometric analysis of shell coiling: coiling in Ammonoids. Journal of Paleontology, 41, 43-65. — and sepkoski, j. j. 1982. Mass extinctions in the marine fossil record. Science, 215, 1501-1503. rich, d. 1989. Image storage has geologic applications. Geotimes , 34, 10-1 1. 112 PALAEONTOLOGY, VOLUME 40 sepkoski, J. J. 1993. Ten years in the library: new data confirm paleontological patterns. Paleobiology, 19, smith, p. L. 1986. The implications of data base management systems to paleontology: a discussion of Jurassic ammonoid data. Journal of Paleontology, 60, 327-340. — tipper, h. w., taylor, D. G. and guex, J. 1988. An ammonite zonation for the Lower Jurassic of Canada and the United States: the Pliensbachian. Canadian Journal of Earth Sciences, 25, 1503-1523. younker, J. l. and ehrlich, R. 1977. Fourier biometrics: harmonic amplitude as multivariate shape descriptors. Systematic Zoology, 26, 336-346. 43-51. BO LIANG PAUL L. SMITH Earth and Ocean Sciences University of British Columbia Typescript received 1 February 1996 Revised typescript received 29 July 1996 6339 Stores Road Vancouver, B.C. Canada, V6T 1Z4 FUNCTIONAL SIGNIFICANCE OF THE SPINES OF THE ORDOVICIAN LINGULATE BRACHIOPOD ACANTHAMBONIA by ANTHONY D. WRIGHT and JAAK NOLVAK Abstract. A giraffid skull and mandible from the early Mid Miocene Keramaria Formation at Thymania (Island of Chios, Greece) has enabled revision of the genus Georgiomeryx. The new specimen is compared with attachment spines, supplementing a pedicle which is functional throughout ontogeny and regarded as anchoring the animal possibly to algal strands above the sea floor. Apart from rare undersized spines with tapering apices, the bulk of the spines on the shell surface are open-ended, with a length attaining half that of the shell itself. The open distal ends of these thin-walled spines would have housed mantle tissue during life, interpreted as being sensory and substituting for setae in the post larval stages. The alternation of spines along successive laminae indicates their interfingering disposition along the anterior and antero-lateral valve margins, where an additional function would have been to screen out coarse particles from the mantle cavity. The minute but distinctive Ordovician spiny lingulate brachiopod Acanthambonia has been recorded from a dozen or so areas in Europe and North America since it was first established by Cooper (1956). Despite this relatively wide distribution, the individual samples documented are typically of one or two valves or fragments. This paucity of material has impeded any unequivocal assessment regarding the systematic placement of the genus, as noted by Popov and Nolvak (1987). This rarity also characterized most of the East Baltic borehole material examined by them, with the sole exception of the Estonian Yiljandi borehole from which over one hundred specimens were obtained from close to the base of the Dicellograptus complanatus graptolite Biozone. This material was particularly useful in that, with the high quality preservation, a minute external pedicle foramen and internal pedicle tube were observed on the valve previously regarded as being the dorsal valve. This contributed to the definite placement of the spine-bearing genus with the typically spiny Siphonotretoidea (Popov and Nolvak 1987). The present paper examines the best preserved specimen from the Viljandi borehole, obtained from a small sample immediately above the base of the Pirgu Stage (sample 14 of Popov et al. 1994, text-fig. 2). This specimen is unique for the genus in the outstanding preservation of its spines which are half as long as the valve itself (Text-fig. 1a). This is proportionately similar in length to those described by Grant (1966) for the productoid Waagenoconcha , but of a totally different order of magnitude as regards absolute size. This raises the intriguing question of the function of such extraordinarily long spines in such a minute brachiopod. Figured Specimen. Acanthambonia portranensis Wright, 1963. Ventral valve, IGT Br 1313: Viljandi borehole, Pirgu Stage, Sample 14 from depth 31T3m (see Popov et al. 1994, text-fig. 2). Repository - Institute of Geology, Tallinn, Estonia. Dimensions. Length — 119 mm ; width —1-21 mm; height - 392 pm; diameter of foramen - 32-3 pm; maximum preserved length of spine - 589 pm. IPalaeontologv, Vol. 40, Part 1, 1997, pp. 113-1 19| © The Palaeontological Association 114 PALAEONTOLOGY, VOLUME 40 text-fig. 1 . Acanthambonia portranensis Wright, 1963. Viljandi borehole, Estonia ; Ordovician (Ashgill), Pirgu Stage (Flc). Ventral valve exterior (IGT Br 1313). a, c, e, direct view of valve, x 50, and enlargements to show spine detail, x 100 and x 200; b, detail of posterior view of valve, x 370; d, f, anterior view of spines, x 200, and detail, x 400. THE FUNCTIONS OF BRACHIOPOD SPINES Spines of many forms occur in different groups of animals and have many, indeed often several, functions varying from defence (in porcupines) to attack (in rhinoceroses), and from locomotion (in echinoids) to food trapping surfaces (in planktonic foraminifera). In the Brachiopoda, Williams and Rowell (1965, p. H84) noted two basic kinds of spine; those formed on a lamellose shell surface by radial ribs projecting away from the shell surface and having WRIGHT AND NOLVAK: ORDOVICIAN BRACHIOPOD 115 their sides joined beneath (such as in Spinorthis and Tegulorhynchia ), and those which open onto the shell interior and characterize the productidines, chonetidines, occasional rhynchonellides such as Acanthothiris, and the siphonotretides. Of these, the largest group is the Productidina, for which the distribution and formation of the spines were discussed in some detail by Muir-Wood and Cooper (1960). The productidine spines were variously employed for direct anchoring in attached forms, for example in reefal environments, or for balancing in the case of forms living free on or partially buried in soft sediment. Other forms were attached by clasping spines in young stages but lay loose on the sea floor as adults. As well as attachment, fine spines, particularly when present on dorsal valves, have been considered to be protective by discouraging epifaunal settlement, the weight of which would have impeded valve opening. Marginal spines (as in Chonosteges ) would have acted as a strainer for the water entering the mantle cavity. Whatever the subsequent relationships to the substrate, in the same way that all bivalves have an initial byssal attachment, so all brachiopods attach in the larval stages for stability even if the peduncular structure atrophies before the adult stage is reached. Grant (1963) demonstrated the attachment of a species of Linoproductus by spines restricted to the posterior margin of the ventral valve and which grew medianly in an arc to hook round the stem of a crinoid. An entire ring could be formed fortuitously if the epithelia in the apices of two oppositely directed spines happened to meet and fuse, but for the most part the individual spines simply hook around the cylindrical object. The fusion is interesting as it confirms that the distal ends of the spines were of soft tissue during growth so that the characteristically hollow open ends are not necessarily the result of breakage. For Waagenoconcha , Grant (1966) showed that the convergent attachment spines of the juvenile stages were replaced in the adult by numerous long, thin spines which served to anchor the animal within the substrate. A third abrupt change in this stock was to smaller spines along the edges of mature ventral valves. These were interpreted as projecting above the substrate, and indicated some function other than that of shell support. The hollow nature of this type of spine passing through to the interior and lined with epithelium suggested to Williams and Rowell (1965, p. H84) that the apical cells could have served to secrete a chitinous pad for attachment to the substrate, with growth of the spine ceasing should the inner end be sealed over by subsequent shell deposition. Rudwick (1965), with particular reference to Acanthothiris , suggested that the tips of the hollow spines could have borne highly sensitive mantle edge tissue which would have functioned in a sensory capacity well in front of the commissure. In this case the apical cells would have had a seta-like sensory function as long as the tips remained open and the internal contact with the mantle was maintained. A half-way stage towards this envisaged spinal function is seen in the numerous orthide stocks which possess aditicules along the rib crests. These are interpreted as accommodating setae within the body of the shell (Wright 1981), and which again maintained contact with the mantle before being sealed oft' internally, sometimes well behind the shell margin (Wright and Rubel 1996, pi. 1). Setae may also have occupied the pits along the posterior margin of the plectambonitoid Eochonetes in the living animal (Wright 1996), although the function here is interpreted primarily as one of balancing, as suggested by Brunton (1972, p. 23) for the similarly disposed spines of the chonetidines. In addition to the sensory function of the spines in Acanthothiris , Rudwick (1965, p. 607) also noted that the spines along the commissure projected radially, alternating in position on each valve to provide an additional function as a protective grille for restricting the size of particles entering the mantle cavity as noted above for Chonosteges and recorded for a number of other articulate stocks. In the lingulates, hollow spines are typical of the Siphonotretoidea, the order which now includes Acanthambonia as discussed above. The spines may be rather sparsely scattered as in Helmersenia (Rowell 1962, pi. 30, fig. 27), but are commonly aligned along growth laminae as in Nushbiella Popov. The spine-bearing lamellae are well displayed in N. lillianae (Holmer 1989, p. 162). Here the edge of each lamella bears a single row of hollow tapering spines in which larger spines alternate with smaller ones; these again alternate in position in successive rows to produce a fine meshed array. At the front, the rows of spines, extending parallel to the valve surfaces, are depressed across 116 PALAEONTOLOGY, VOLUME 40 the commissure. With the convexity of the valves, the spines in this position would have formed an interlocking grill during life as envisaged by Rudwick (1965) for Acanthothiris. If these forms with hollow spines around the commissure contained sensory epithelium in their apices, as appears likely, and if the spines were thus deployed as a sensory array well in front of the shell, there would be little point and indeed little space for the simultaneous development of marginal setae. It therefore seems reasonable to suggest that these forms would not have possessed setae in the adult shells. Although setae are almost ubiquitous in Recent brachiopods, modern adult Neocrania lacks setae; presumably the extinct spinose Acanthocrania also lacked them. Lacazella is another living stock which is also known to lack setae in the adult, so although the presence of setae is well known from Cambrian lingulates, especially from the Burgess Shale, it is also reasonable to accept that the shedding of post larval setae may well have taken place in those late Cambrian and Ordovician forms with commissural spinal arrays. THE SPINES IN AC ANT H A M BO N I A The spines of Acanthambonia are typical of the hollow spines in several brachiopod groups which open on the shell interior, and would have developed in the manner indicated by Williams and Rowell (1965, p. H84). Thus the spines would have been lined with outer epithelium proliferated by the apical cells of generative tips secreting the usual sequence of organic and inorganic layers as the spines continued to grow. In Acanthambonia there are fairly persistent pits on the inner surface which correspond to the spines on the exterior, as seen in Popov and Nolvak (1987, pi. 2, fig. la); these indicate that contact continued to be maintained during growth between the outer epithelium lining the spine and that lining the shell interior. Development and function At the posterior apex of the larval shell is situated the pedicle foramen (Text-fig. 2d). In this specimen, and in all the specimens studied by Popov and Nolvak (1987), the foramen and internal pedicle tube remained open throughout all stages of ontogeny. Thus the pedicle was functional throughout life even though its size remained constant as the shell grew. The median position of the pedicle tube is reflected on the surface of the smooth larval shell and defined laterally by a pair of depressions (Text-fig. 2e). With growth, fine concentric fila, commonly broken into drapes by nick points (Text-fig. 2g) as described for acrotretoids generally by Williams and Holmer (1992), are succeeded by five or six coarse, irregular but spineless concentric folds (Text-fig. 2f). The first sub- erect and anteriorly directed spine then appears at the front followed by another (Text-fig. 2f) before the first pair of spines developed on the posterior margin. The latter, together with the subsequent spines along the posterior margin, are curved medianly towards the umbo (Text-fig. 1a) in the manner of productidine clasping spines. Even allowing for breakage their shortness and moreover their stronger curvature, in which they hook over antero-medianly (Text-fig. 1b), indicate that they could not have formed encircling spines as noted for Linoproductus (Grant 1966), Plicatifera and other forms (Brunton 1966). The hook-like appearance does, however, suggest a function of attachment, and it may be that they served to provide ancillary support for the pedicle attachment of the growing shell by hooking over algal strands, to glue on to an attachment surface or into sediment. If the hooks were related to algal strands, their disposition would suggest that the hinge lay parallel to the growth direction of the alga and essentially perpendicular to the sea floor if the algae were benthic. Alternatively, if the tissue-filled distal ends remained sensitive, the curvature of these spines may simply reflect a response away from a surface to which the shell was attached by the pedicle. The possibility of attachment of Acanthambonia to algae is supported by the convincing demonstration of Havhcek et al. (1993) for such attachment by the larger lingulid Rafanoglossa and also small fossil articulates. However, an alternative attachment surface would be that of sponge spines, as demonstrated by the attachment of Dictyonina to Pirania in the Burgess Shale fauna (Whittington in Conway Morris et al. 1982, p. 25, pi. R). In the case of WRIGHT AND NOLVAK: ORDOVICIAN BRACHIOPOD 117 microbrachiopods, the attachment scar of the Eoconulus can take the form of smooth groove, suggesting attachment to a very fine cylindrical object (Wright and McClean 1991, p. 125) which could well be a sponge spine. A sponge substrate would certainly have been available for the Acanthambonia, as the association of sponge spicules with the lingulates in the Estonian faunas has already been noted by Popov et al. (1994, p. 628). Away from the posterior margins, the increasing valve size resulted in an increasing number of evenly spaced spines being developed on the growth lamellae. These extended perpendicularly to the margin of the time and built up an array, with the spines of each lamina being offset with regard to those of the preceding and succeeding laminae (Text-fig. 1). When viewed directly (Text-fig. 1 e) these spines appear straight, but some irregularity in growth is clear when viewed anteriorly (Text-fig. Id, f) or laterally (Text-fig. 2b). Two of the spines, the seventh and seventeenth from the text-fig. 2. Acanthambonia portranensis Wright, 1963. Viljandi borehole, Estonia ; Ordovician (Ashgill), Pirgu Stage (Flc). Ventral valve exterior (IGT Br 1313). a-b, posterior and lateral views of valve, x 65; c, broken outer layer showing opening through inner layer below, x 810; d-g, posterior and anterior views of larval shell (d, f), x 370, x 400, and details (e, g), x 1600, x 1600. 118 PALAEONTOLOGY, VOLUME 40 left in Text-figure lc, and also in Text-figure 1d-f, have sealed tips. The latter has a particularly narrow base and suggests that growth was inhibited by its smaller number of secreting cells with consequent narrowing of the diameter and distal closure. A small spine that is distally sealed can also be observed in a specimen of Cyrbasiotreta figured by Williams and Curry (1985, fig. 56b), another form in which the edge of each lamella possessed a single row of fine, evenly spaced hollow spines. These examples raise the question of whether distal tapering and sealing was the norm, and that those with open ends are simply reflecting breakage of the fine points. Certainly the spines must have remained distally open during growth otherwise there would have been no growth. When spines go out of commission through proximal sealing, as in the case also of setae embedded in the shell as well as in productide and Acanthothiris spines, the bulk of the envisaged sensory function would have been lost as the isolated tissue within the spine withered. But the Acanthambonia ventral interior figured by Popov and Nolvak (1987, pi. 2, fig. la-b) suggests that sealing at the inner ends of the spines is less common in this stock. In such cases there seems to be no necessity for a closure at the distal end. The longest spines occur at the front of the specimen, which would perhaps be expected as the shell increments here are greatest. For the Bohemian species A. klabavensis, Havhcek (1982, p. 74) noted that the longest spines were directed laterally. The view of the entire Estonian shell (Text-fig. 1a) suggests that the spines increased in length around the shell margin from posterior to anterior, and also with the increase in shell size. Impressive though the spines are, some are broken and indeed a broken spine is still lying across the surface. The spines are shortest in the postero-central part of the valve where a few are reduced almost to the base. This presumably is a result of a taphonomic effect rather than abrasion from the valve rubbing against a surface during life, as a similar shortening is to be seen on a dorsal exterior (Popov and Nolvak, pi. 1, fig. la-b). It is noticeable in the Acanthambonia that the internal diameter of the spine distally is relatively large (Text-fig. Id, f); there is no evidence of the wall of the spine becoming thicker as in the typical productidine spine. In the latter case, the more robust the spine becomes, the better it is for the supporting function; the thinner, non-thickening wall in Acanthambonia allows the maximum amount of tissue internally which would be important to the sensory function, so on this criterion the Acanthambonia spines are here interpreted primarily as sensory. The alternating arrangement of the long spines around the anterior and antero-lateral parts of the shell suggests that these long mature spines also functioned to screen coarser particles from entering the mantle cavity in the manner envisaged for the very much larger articulate Acanthothiris by Rudwick (1965). A supportive role appears to have applied only to the spines along the posterior margin. With a functional pedicle throughout ontogeny, attachment, probably to algal strands above the sea floor, is assured; the minute size would have mitigated against a sedentary position directly on a mud surface. Further, the similar distribution of spines on both valve surfaces would be unlikely were one to be continually adjacent to or partially submerged in sediment; this again suggests that both valves were in immediate contact with the water. Acknowledgements. We are grateful to Dr Cyprian Kulicki (Warsaw), who kindly gave valuable assistance with the preparation of the SEM images; and to Dr David Harper (Galway) for his helpful comments on the draft manuscript. We are indebted also to the referee. Dr Lars Holmer (Uppsala), for making positive suggestions for the improvement of the manuscript. REFERENCES brunton, c. h. c. 1966. Silicified brachiopods from the Visean of County Fermanagh. Bulletin of the British Museum ( Natural History ), Geology Series, 12, 173-243. — 1972. The shell structure of chonetacean brachiopods and their ancestors. Bulletin of the British Museum ( Natural History), Geology Series, 21, 1-26. cooper, g. a. 1956. Chazyan and related brachiopods. Smithsonian Miscellaneous Collections, 127, parts 1 and 2, 1-1245. WRIGHT AND NOLVAK: ORDOVICIAN BRACHIOPOD 119 CONWAY MORRIS, S., WHITTINGTON, H. B., BRIGGS, D. E. G., HUGHES, C. P. and BRUTON, D. L. 1982. Atlas of the Burgess Shale. Palaeontological Association, London, 31 pp. grant, r. e. 1963. Unusual attachment of a Permian linoproductid brachiopod. Journal of Paleontology , 37, 134-140. — 1966. Spine arrangement and life habits of the productoid brachiopod Waagenoconcha. Journal of Paleontology , 40. 1063-1069. havlicek, v. 1982. Lingulacea, Paterinacea, and Siphonotretacea (Brachiopoda) in the Lower Ordovician sequence of Bohemia. Sborntk geologicky ved Paleontologie , 25, 9-82. — vanek, J. and fatka, o. 1993. Floating algae of the genus Krejciella as probable hosts of epiplanktic organisms (Dobrotiva Series, Ordovician; Prague basin). Journal of the Czech Geological Society , 38, 79-88. holmer, L. E. 1989. Middle Ordovician phosphatic inarticulate brachiopods from Vastergotland and Dalarna, Sweden. Fossils and Strata , 26, 1-172. muir-wood, h. M. and cooper, G. a. 1960. Morphology, classification and life habits of the Productoidea (Brachiopoda). Memoirs of the Geological Society of America , 81, 1-447. popov, L. and nolvak, j. 1987. Revision of the morphology and systematic position of the genus Acanthambonia (Brachiopoda, Inarticulata). Eesti NSV Teaduste Akadeemia Toimetised , Geoloogia, 36, 14-19. — and holmer, l. E. 1994. Late Ordovician lingulate brachiopods from Estonia. Palaeontology , 37, 627-650. rowell, a. j. 1962. The genera of the brachiopod superfamilies Obolellacea and Sipohonotretacea. Journal of Paleontology , 36, 136-152. rudwick, m. j. s. 1965. Sensory spines in the Jurassic brachiopod Acanthothiris. Palaeontology , 8, 604—617. williams, a. and curry, g. b. 1985. Lower Ordovician Brachiopoda from the Tourmakeady Limestone, Co. Mayo. Bulletin of the British Museum ( Natural History ), Geology Series , 38, 183-269. and rowell, a. j. 1965. Morphology. 57-155. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part H. Brachiopoda. Volume I. Geological Society of America and University of Kansas Press, Lawrence, Kansas, 521 pp. — and holmer, l. e. Ornamentation and shell structure of acrotretoid brachiopods. Palaeontology , 35, 657-692. wright, a. d. 1963. The fauna of the Portrane Limestone. 1. The inarticulate brachiopods. Bulletin of the British Museum (Natural History ), Geology Series , 8, 221-254. - 1981. The external surface of Dictyonella and other pitted brachiopods. Palaeontology , 24, 443-481. — 1996. The taxonomic importance of body-mantle relationships in the Brachiopoda. 299-304. In copper, P. and TISUO jin (eds). Brachiopods. Proceedings of the Third International Brachiopod Congress, Sudbury , Ontario, Canada, 2-5 September 1995. — and McCLEAN, A. E. 1991. Microbrachiopods and the end-Ordovician event. Historical Biology , 5, 123-129. — and rubel, M 1996. A review of the morphological features affecting the classification of clitambonitidine brachiopods. Palaeontology, 39, 53-75. Typescript received 27 March 1996 Revised typescript received 1 July 1996 ANTHONY D. WRIGHT School of Geosciences The Queen’s University Belfast BT7 INN Northern Ireland JAAK NOLVAK Geoloogia Instituut Estonia Puiestee 7 Tallinn EE0105 Estonia A GIRAFFID FROM THE MIDDLE MIOCENE OF THE ISLAND OF CHIOS, GREECE by louis de bonis, george d. koufos and sevket sen Abstract. A giraffid skull and mandible from the early Mid Miocene Keramaria Formation at Thymania (Island of Chios, Greece) has enabled revision of the genus Georgiomeryx. The new specimen is compared with Canthumeryx syrtensis from Libya and Injanatherium from Iraq and Arabia. It is concluded that all these genera, which have flat and laterally directed horns, belong, together with Giraffokeryx , to the subfamily Canthumerycinae. New palaeomagnetic data from Chios combined with the presence of the fossil mammals provide new evidence for dating the locality to chron C5Br or early MN5. Since 1940, when Paraskevaidis described some mammalian remains from Chios, several opinions referring to the composition, the relationships and the age of the fauna have been expressed. The initial scanty and fragmentary material was not sufficient to provide firm conclusions. During the 1960s, a further collection was made in which the most significant discovery was a complete skull and mandible of a mastodont (Melentis and Tobien 1967; Tobien 1980). Other mammalian remains were very fragmentary and belong mainly to bovids and giraffids (Lehmann and Tobien in press) although their identifications are tentative. Some micromammalian remains were also noted in a preliminary faunal list (Tobien 1968), but they have never been described. A new study of the bio- and magnetostratigraphy of the Neogene deposits of Chios has been carried out by a Hellenic-French team during 1991 and 1993. The material unearthed during this includes a rich micromammalian fauna and a few, quite complete macromammalian remains. The mammalian localities are in the Keramaria Formation, which consists of sands and silts alternating with siltstones and sandstones (for more details see Kondopoulou et al. 1993 and Koufos et al. 1995). All the macromammalian remains were found in the level ‘Thymiana B' (THB). They include a piece of skull and a left mandible with most of the cheek teeth of the same giraffid. The first remains of a giraffid from the Neogene deposits of Chios were described by Paraskevaidis (1940) under the new binomen Georgiomeryx georgalasi on the basis of a mandibular fragment with P2/-P3/ (Paraskevaidis 1940, pi. 13, figs 4—5). The comparison of such poor material with other taxa was very difficult, and for a long time its taxonomic position was unknown. The new material is more complete and increases our knowledge of the anatomy, the phylogenetic relationships and the systematics of this Miocene giraffid. SYSTEMATIC PALAEONTOLOGY Family giraffidae Gray, 1821 Genus georgiomeryx Paraskevaidis, 1940 Type-species. G. georgalasi Paraskevaidis, 1940. Diagnosis. As for the type and only species. [Palaeontology, Vol. 40, Part 1, 1997, pp. 121— 133| © The Palaeontological Association 122 PALAEONTOLOGY, VOLUME 40 Georgiomeryx georgalasi Paraskevaidis, 1940 Text-figures 1-3, 4c Holotype. Mandibular fragment with P/2-P/3 (Paraskevaidis 1940, pi. 13, figs 4—5) housed in the Paraskevaidis collection, Athens. Paratype. Skull and left mandible (THB-30, 16), housed in the University of Thessaloniki, Laboratory of Geology and Paleontology. Locality and Horizon. Thymiana B, Chios, Greece. Keramaria Formation, Middle Miocene (MN 5). Diagnosis. Primitive giraffid with a pair of flat and laterally oriented horns, situated just over the orbits. Brachyodont. Large upper premolar and molar cingulum. P3/ and P4/ heteromorphic. Differences between Georgiomeryx georgalasi and related genera are as follows. Canthumeryx differs from G. georgalasi by its less molarized lower premolars. Injanatherium differs from G. georgalasi by the presence of an anterior pair of horns, the second pair (posterior) situated back to the orbits, and in its more advanced dentition (lesser brachyodonty, weaker or absent upper cheek tooth cingulum). Giraffokeryx differs from G. georgalasi in the presence of an anterior pair of horns, a second pair situated back to the orbits, in the less lateral direction of the horns, and in the more advanced dentition (semi-hypsodonty, weaker or absent cingulum). DESCRIPTION Skull The new skull is crushed, and both anterior and posterior portions are lacking. Its dimensions suggest a medium-sized giraffoid. The anterior part of the frontal is partially broken but there is no trace of an anterior pair of horns or ossicones, the remains of which would have been apparent as a bump on the frontal bone if they were present (as is the case for Giraffokeryx). On the left side of the skull, one of a posterior pair of horns is located exactly above the orbits and slightly posteriorly over the temporal fossa. The horn is flat and laterally oriented (Text-figs 1-2). A temporal crest starts from each horn on the parietal and passes toward the back of text-fig. 1. Georgiomeryx georgalasi Paraskevaidis, 1940. Skull and mandible (THB 30) in lateral view; Thymiana, MN 5 (lower Middle Miocene), Chios, Greece, Scale bar represents 30 mm. de BONIS ET AL.: MIOCENE GIRAFFID 123 text-fig. 2. Georgiomeryx georgalasi Paraskevaidis, 1940. Skull (THB 16) in dorsal view; Thymiana, MN 5 (lower Middle Miocene), Chios, Greece. Scale bar represents 30 mm. the skull which is lacking. Despite the crushing of the skull we can see that the maxilla is deep. The orbit is large, rounded, without any trace of a lacrimal foramen. The zygomatic arch, well preserved on the left side, is weak and not projected (Text-fig. 1). The palate is narrow in front of P2/ sockets and it widens posteriorly. The glenoid surfaces are widened laterally and they are more convex than in modern giraffes. Mandible The corpus is long and shallow especially forward of the dentition as in typical giraffes. The ramus is partially broken (Text-fig. 1). Dentition The dentition is characterized by the rugose structure of enamel, a giraffid characteristic. Upper dentition (Text-fig. 3). All the teeth are very low. P2/ is missing on both sides. P3/ is widened posteriorly and has a triangular occlusal profile. The lingual cusp (‘protocone’) is elongated and crescentoid; it has an anterior vertical lingual groove and a well-marked spur on its buccal face. The buccal wall of the crown is clearly asymmetrical; the anterior part is smaller than the distal one and both are separated by a strong rib. The parastyle is sharp and well developed while the metastyle, also well developed, is more rounded. There is a slight lingual cingulum. P4/ is more symmetrical than P3/ and its length is smaller than its breadth. The external rib is central on the buccal wall. The buccal cusp (paracone) has also a vertical rib on its lingual face. There is a very large lingual cingulum which continues on the lingual part of the mesial face and on the entire 124 PALAEONTOLOGY, VOLUME 40 text-fig. 3. For caption see opposite de BONIS ET AL. \ MIOCENE GIRAFFID 125 table 1. Measurements (in mm) of the upper teeth of Georgiomeryx georgalasi: Thymiana (Chios); Middle Miocene (MN5). LP3/ LP4/ LM1/ LM2/ LM3/ bP3/ bP4/ bMl/ THB-30 19-6 170 22-0 23-3 24-1 16-1 19-7 21-5 bM2/ bM3/ b/L P3/ b/L P4/ b/L Ml/ b/L M2/ b/L M3/ LM1/-M3/ 24-1 23-6 82-0 116-0 97-7 1030 97-9 67-8 table 2. Measurements (in mm) of the lower teeth of Georgiomeryx Miocene (MN5). georgalasi ; Thymiana (Chios); Middle LP/3 LP/4 LM/1 LM/2 LM/3 bP/3 bP/4 bM/1 THB-16 19-2 bM/2 14-5 21-0 bM/3 14-8 21-3 b/L P/3 47-11 22-2 b/L P/4 52-0 31 b/L M/1 60-4 9-2 b/L M/2 65 3 11-0 b/L M/3 47-7 12-7 LM/l-M/3 76-7 distal face where the wear has caused a post-fossette. The size of the molars increases from Ml/ to M3/ (Table 1 ). Ml / is broken on both sides of the maxilla but on the right side there is a weak metacone rib and a reduced metastyle. On M2/ and M3/ the styles are strong and there is a buccal cingulum and a paracone rib. The mesial and lingual cingulum is well marked, especially around the protocone of M3/. The distal part of the last molar is not reduced as in modern giraffes. There is no entostyle. Lower dentition (Text-fig. 3). Like their upper counterparts, the lower cheek teeth are very low. All have relatively narrow and elongated crowns (Table 2). P/2 is broken. P/3 has five transverse crests and it is not molarized although the median crest is thickened on the lingual side, and the posterior valley is closed on the buccal side. P/4 is a typical giraffid premolar. The anterior valley is closed by a metaconid, the entoconid is well developed, the posterior valley is open on both buccal and lingual sides, hypoconid and posterior transverse crest are clearly separated from the rest of the crown. On the molars, the metastylid and the entoconulid are weak on M/1 and more developed on M/2 and M/3. There is no lingual rib except a very weak one on M/3. The M/3 hypoconulid is very twisted toward the buccal side. An ectostylid is present on M/1, weakly developed on M/2 and even more so on M/3. There is no trace of any Palaeomeryx- fold. COMPARISONS Skull Two characters of the Chios skull, absence of preorbital fossa and absence of lacrimal foramen, are currently considered as synapomorphies of Giraffidae and Giraffinae respectively, so we will restrict the comparisons to the giraffids. Giraflfokeryx. This genus was erected by Pilgrim (1910) for the species G. punjabiensis. The material, some fragmentary and scattered dentitions coming from the lower Siwaliks of the Salt Range and from localities near Chinji, was described and figured later (Pilgrim 191 1). A skull unearthed from the middle Siwaliks of northern Punjab has been referred to Giraffokeryx (Colbert 1933). All these fossiliferous localities are of mid Miocene age (Pilgrim 1934). The skull of Giraffokeryx has two text-fig. 3. Georgiomeryx georgalasi Paraskevaidis, 1940. A, right upper cheek teeth in occlusal view; b, left upper cheek teeth in occlusal view; c, left lower cheek teeth in occlusal view; D, right upper cheek teeth in buccal view; E, left lower cheek teeth in lingual view; F, left lower cheek teeth in buccal view. Scale bar represents 20 mn. 126 PALAEONTOLOGY, VOLUME 40 pairs of horns. The anterior pair is situated on the anterior part of the frontal, clearly in front of the orbits ‘above the first and second molars’ and ‘they are confluent at their bases’. The posterior horns arise ‘directly back of the orbits’, their bases being fronto-parietal. All these horns are directed upward, backward and laterally. This pattern does not correspond to the Chios skull, the horns of which arise directly over the orbits and are more laterally oriented. Palaeotragus. The known skulls of Palaeotragus (e.g. Hamilton 1978; Geraads 1986) are characterized by a pair of simple supraorbital horns directed upwards, laterally and slightly backwards. These horns are, however, inclined slightly backwards. They have a rounded or oval basal section. This is the same for an isolated horn described under the name Propalaeotragus actaensis Godina, Vislobokova and Abdrachmanova, 1993 from the Miocene of Kazakhstan. The females can be hornless (Colbert 1936). Canthumeryx ( = Zarafa). An edentulous skull coming from the lower Miocene of Gebel Zelten (Libya), holotype of the species Zarafa zelteni (Hamilton, 1973), has been synonymized with Canthumeryx syrtensis Hamilton, 1973 which had been described from two mandibular fragments and an isolated M/3 from the same locality. Upper and lower teeth allocated to Zarafa are now identified also as C. syrtensis. The Canthumeryx skull is broken anteriorly and posteriorly as is the Chios skull but the damage is far less on the North African skull. In dorsal view, the frontal is flat with a pair of laterally expanded flat horns with triangular bases very similar to that of Chios skull. These horns are also located just over the orbits and Canthumeryx is quite similar to the Chios skull. Injanatherium. Erected for a toothless skull from late Miocene Gebel Hamrin layers (Iraq), type specimen of the species I. hazimi Heintz, Brunet and Sen, 1981, Injanatherium has been also found in Saudi Arabia in the mid Miocene Hofuf Formation near A1 Jadidah (Morales et al. 1987) with I. arabicum. Injanatherium is characterized by laterally horizontally extended and relatively supra- orbital horns. A large pair of robust horns is situated above or a little behind the orbits, over the temporal fossa. It is impossible to know whether I. hazimi had a second anterior pair of horns but a fragment of skull from Al Jadidah has a small horn in front of the orbit just posterior to the fronto-nasal suture. It has been suggested that a piece of skull identified as Samotherium sinense (Bohlin, 1926, text-fig. 137) could belong to the same group. Injanatherium is similar to the Chios skull and Canthumeryx in its possession of lateral horizontal horns. However, the Al Jadidah remains show that there is an anterior pair of horns and that the posterior one is less flat, more elongated and set a little further back from the orbit without turning upwards. Palaeomerycidae. The skull of Palaeomeryx is especially well known from specimens from the middle Miocene of China (Qiu et al. 1985), and that of Triceromeryx from fragmentary specimens from Spain (Crusafont-Pairo 1952; Astibia and Morales 1987). These two genera have an elongated skull with three horns. Two horns are over the orbits and the third, median, is on the occipital. They differ from the Chios skull in the shape of the orbital horns which are vertical and by the presence of a lacrimal foramen and a pre-orbital fossa. Sivatheriinae. The species belonging to this group are characterized (Geraads 1986) by large size and by the cranial appendages behind the orbits, except in the females which can be hornless. These characters do not correspond to those of the Chios skull. Dentition Giraffokeryx. According to Pilgrim (1911), the first described specimens of G. punjabiensis come from ‘one small spot near Phadial ... which were remarkably free from the remains of any other animal except an antelope’; all the material can probably be referred to the same species. The upper de BONIS ET AL.: MIOCENE GIRAFFID 127 cheek teeth differ from those of the Chios skull in the absence of a cingulum although P3/ and P4/ have the same general shape. The Giraffokeryx molars have a spur in the anterior fossette which is not present on Chios molars. The lower premolars, P/3 and P/4, have the same general pattern as have the Chios premolars but they are less narrow and more hypsodont, P/3 has a more pronounced buccal groove, its anterior portion is less elongated and the transverse crests are less oblique. P/4 has also a more reduced anterior portion. The lower molars, M/1 and M/2, have different proportions (Table 2). The metastylid of M/3 is less developed. The M/3 hypoconid is less twisted and the ectolophid is more developed. Like the premolars, the molars are also more hypsodont, even taking into account the dental wear. For Pilgrim (1911) the relative hypsodonty was one of the main characters of Giraffokeryx. The skull described by Colbert (1933) seems to have the same characters as Pilgrim’s specimens. The locality of Prebreza (Serbia) has yielded some dental remains identified as Giraffokeryx (Ciric and Thenius 1959; Pavlovic 1969). The upper tooth material consists of left and right maxillae with D3/-M 1 /, right P2/-P4/ belonging to an older individual and a broken piece with M2/. Compared with the Chios teeth, P3 / is more symmetrical, more giraffe-like, P4/ has a similar pattern but with a smaller breadth, a weaker buccal rib and absence of lingual cingulum. The upper molars seem to have a slightly higher degree of hypsodonty, a less marked paracone buccal rib and a weaker cingulum. The lower teeth are known from a piece of mandible with P/2-M/3 and two other ones with M/2-M/3. P/3 has the same pattern but the transverse crests are less oblique. The P/4 talonid breadth is larger and, perhaps due to differential wear, the posterior transverse crest is not isolated. The comparison of M/2-M/3 is more puzzling. They have the same degree of hypsodonty for one specimen (Pavlovic 1969, pi. 15) and a higher degree on the second one (Pavlovic 1969, pi. 16). M/3 has a better marked ectostylid and a weaker metastylid. A few remains (P2/-P/4 and P/3-P/4) from the Upper Miocene of Nakali have been attributed to Giraffokeryx (Aguirre and Leakey 1974). These teeth are larger than the Chios teeth. P3/ is elongated but more symmetrical without any vertical lingual groove. The P4/ cingulum is weaker. P/3 is more evolved, more giraffe-like with a larger talonid. A left mandibular corpus and several isolated teeth from the mid Miocene layers of Pasalar (Turkey) have been identified under the name Giraffokeryx aft', punjabiensis (Gentry 1990). The premolars display a large amount of variation. Thus two teeth are quite similar to the Chios P/3 (Gentry 1990, fig. 5g-h) but they are DP/3 and another one is more massive with a larger breadth (fig. 5f). P/4 could be very evolved (fig. 5d ) or very primitive, Palaeomeryx-\\kt (fig. 5c); it can have a large breadth (fig. 5a) or be narrow (fig. 5b). We consider the Pasalar material to be too scarce and incomplete to permit any precise identification. It may comprise more than one taxon. Palaeotragus. This genus appears to have existed from the mid to the late Miocene but the different species allocated to it can display some significant differences, and it may prove to be paraphyletic (Hamilton 1978). Following Geraads (1986), all the palaeotragines could be included in the same genus except, we presume, Canthumeryx and Giraffokeryx. We consider here the middle Miocene species; the upper Miocene ones (P. rouenii , P. coelophrys and P. germaini) are larger and possess some derived characters lacking on the Chios material. P. primaevus comes from the middle Miocene of Fort Ternan (Churcher 1970, 1978) and some other Kenyan localities (Hamilton 1978). P3/ has a primitive shape but the lingual cusp is more posteriorly situated ; upper premolars and molars do not display any cingulum : M 1 / and M2/ have a well-developed spur on the buccal face of the metaconule. The lower dentition is less brachyodont and the teeth are more robust (Table 2). P/3 has a small metaconid, a trend toward the crown molarization. P/4 is more giraffe-like with an anteriorly widened crown. The lingual surfaces of the lower cheek teeth are more flattened. P. tungurensis Colbert, 1936 comes from the middle Miocene of Central Asia. The teeth differ by their degree of hypsodonty and the shorter premolars relative to the larger molars (Table 2). P3/ is primitive like that of P. primaevus. The upper molars have no cingulum, the mesostyle is higher vis-a-vis the paracone buccal wall and lower vis-a-vis the metacone buccal wall, and the paracone 128 PALAEONTOLOGY, VOLUME 40 rib is weaker. P/3 is more molarized, the anterior portion of P/4 is widened and the third lobe (hypoconulid) of M/3 is more simple. P. lavocati Heintz, 1976 was founded on scarce and fragmentary specimens from the middle Miocene of Beni Mellal (Morocco). It is difficult to characterize this species but the author concluded that the differences between P. lavocati and P. tungurensis are very slight. Most of the noted differences between Chios teeth and P. tungurensis are present in the Beni Mellal specimens. Canthumeryx ( = Zarafa). This genus is known from a single species from the lower middle Miocene of Gebel Zelten (Hamilton 1973) and other African localities (Hamilton 1978). Canthumeryx is brachyodont as is the Chios fossil. P3 / looks primitive with an asymmetrical crown and an antero- lingual groove. P4/ is quite similar to the Chios material but is a little less asymmetrical, wear facets being present on the two buccal crests of the paracone. The lingual cingulum is also a little less developed. Ml/ and M2/ are also quite similar to that of the Chios fossil but the lingual cingulum is interrupted just in front of the cusps; the spur on the buccal face of the metaconule is very marked in some specimens (Hamilton 1973, pi. 5) and less marked on some other ones (Hamilton 1973, pi. 6). In a specimen from Muruorot (Hamilton 1978), P/3 does not differ a lot from the Chios P/3 but P/4 is far more primitive; a small elongated metaconid is present but the anterior valley is buccally open as in some specimens of Palaeomeryx. The few known lower molars do not differ significantly. We note that a lower molar on a piece of mandible (Hamilton 1973, pi. 5, figs 2-3) is larger than the other teeth identified as C. syrtensis or Zarafa zelteni. It differs also in possessing more wrinkled enamel, a higher crown relative to the length, the very weak ectostylid, the weak or absent lingual rib of the lingual cuspids and the less narrow buccal cuspids. This molar belongs certainly to another species of giraffid and may be a Palaeotragus sp. Injanatherium. Some teeth have been allocated to this genus (Morales et al. 1987). P3/ is more symmetrical and the upper molars lack any lingual cingulum. The lower premolars, especially P/4, and the molars are quite similar but the degree of hypsodonty is a little higher. Palaeomerycidae. In this family, P3/ can be as primitive as is the Chios P3/. P4/ and the molars have a lingual cingulum. P/4 is primitive with a buccally open anterior valley but with a quite marked trend to close it, especially on some large specimens from Sansan (Astibia and Morales 1987). The trend toward the closure of the P/4 anterior valley can occur in some other genera which are probably linked to the giraffoids, for example Hispanomeryx (Moya-Sola 1986) or Teruelia (Moya-Sola 1987), or even in large cervids. DISCUSSION The use of horns and ossicones in giraffid taxonomy is accorded different significance by different authors. Bohlin (1926) perceived giraffid cranial appendages as highly variable and thus not giving any useful information on their relationships. As a result, he could put together in the same genus, or even the same species, skulls with upwardly or laterally directed appendages. However, although horns could be present or absent according to the sex in fossil giraffids, observation of recent species of giraffids leads us to consider horns or ossicones to be a good tool for specific or possibly generic taxonomy. Position and shape of ossicones do not display large variation in living species, so we prefer to hypothesize, as do most authors, that frontal appendage number, shape and position are significant in characterizing not only particular species but also genera or particular lineages. Most recent and fossil giraffids, the latter including Palaeotragus , have upwardly and, sometimes, slightly backwardly or laterally directed frontal appendages. However, very few have purely laterally oriented ones with a flattened base. This pattern is found in Canthumeryx and Injanatherium although the latter genus has developed a second anterior pair of lateral horns de BONIS ET A L. : MIOCENE GIRAFFID 129 (Morales et al. 1987). Some authors (Morales et al. 1987; Gentry 1994) have supported the hypothesis that both could be put in a same genus, but the synonymy was not firmly established because of some differences. Most of the characters of the Chios skull correspond well with those of Canthumeryx syrtensis. Its frontal appendages have the same shape and are in a similar position just above and slightly behind the orbit. The two known skulls of Injanatherium display a slightly different pattern, with the horns emerging behind the orbits. There is also another difference, in the presence of an anterior second pair of horns situated between the anterior orbital rim and the fronto-nasal suture. This second pair is obviously absent in Canthumeryx syrtensis as well as in the Chios skull (Text-fig. 4). We consider that the appearance of a second pair of horns is a derived J text-fig. 4. Skulls of Canthumerycinae in dorsal view. A, Injanatherium arabicum; b, Injanatherium hazimi; c, Georgiomeryx georgalasi; d, Canthumeryx syrtensis. Scale bar represents 100 mm. Arrows indicate the anteroposterior extent of the orbits. character. So, Injanatherium is more derived than Canthumeryx and the Chios skull, and we favour the opinion that the two sets must be placed in different genera. As shown above, the Chios and Gebel Zelten specimens match quite well in cranial characters but differ in their tooth morphology. Both are brachyodont but the Canthumeryx P/4 is more primitive and does not reach the ‘giraffid stage’ of evolution with a closed anterior valley and with a lingually open posterior valley. In this respect the Chios dentition is more derived (i.e. more giraffid-like) than 130 PALAEONTOLOGY, VOLUME 40 that of Canthumeryx. If we consider the molarization of the premolars as a shared derived character of Giraffidae, we cannot include the Chios specimen and Canthumeryx in the same lineage, genus or family. However, on the other hand, it seems that premolar molarization is a general trend within giraffoids or even Pecora which can occur in Palaeomerycidae or in Cervidae as well, and which can have appeared several times in giraffid evolution. The horn shape, insertion, and orientation seem to be far more significant for giraffid phyletic and systematic assessments. The Chios and Gebel Zelten specimens could belong to the same lineage, in which the premolars follow the same general giraffid trend toward molarization. Canthumeryx is more primitive and the Chios giraffid more advanced in P/4 molarization. A further evolutionary step, unknown here, would have been P/3 molarization and the homomorphy of upper premolars P3/ and P4/. So, both forms must be included in genera different from Injanatherium. A generic name already exists for the Chios giraffid, namely Georgiomeryx Paraskevaidis, 1940. This genus was established on the species G. georgalasi Paraskevaidis, 1940, the type specimen of which is a fragment of mandible with P/2-P/3 from the same locality, Thymiana, where the new skull was collected. The premolars match those of the new specimen very closely and there is no doubt that both belong to the same species of brachyodont giraffid. Until now, it has been very difficult to compare Georgiomeryx remains with other giraffids because of the incomplete nature of the type specimen, but the new discoveries at the type locality permit the recognition of similarities to the Gebel Zelten material. Thus far, Canthumeryx Hamilton, 1973 could have been a junior synonym of Georgiomeryx, but if so, the species Georgiomeryx georgalasi would share some derived dental features (more molarized P/4) with other genera ( Injanatherium and Giraffokeryx ) and the genus Georgiomeryx would be paraphyletic. Until now, Georgiomeryx georgalasi and Canthumeryx syrtensis have been known from Chios and Gebel Zelten from the type material of the two species. Some specimens from Muruarot (Kenya) have been also referred to syrtensis (Hamilton 1978). The two described species represent two successive evolutionary steps, C. syrtensis being the more primitive. Injanatherium is more derived with a second pair of lateral horns (anterior ones), the posterior horn back to the orbits, and the lesser degree of brachyodonty ; Giraffokeryx is differently derived. All these genera constitute together an early offshoot of the family Giraffidae, an offshoot which requires subfamilial , text-fig. 5. Cladogram of the sub-family Can- thumerycinae. Character-states defining the nodes are as follows. 1. Bilobed lower canines; wrinkled enamel. 2. Flat and laterally directed supra-orbital horns. 3. Molarized P/4. 4. One pair of lateral horns slightly posterior to the orbits; one pair of anterior horns; P3/ relatively symmetrical; molarized P/3; relative hypsodonty. 4a. Very robust supra-orbital horns; 4b. Upwardly directed distal horns. distinction (Text-fig. 5). The Gebel Zelten locality, from which C. syrtensis was collected, has been dated to different geological or biochronological levels. It is attributed to the Burdigalian (upper Lower Miocene) by Arambourg and Magnier (1961) and Savage (in Selley 1969). Savage (1989) gives a more precise dating of MN3b. In East Africa, C. syrtensis occurs in Murourot (upper Lower Miocene, ‘set II’) and, with uncertainty, in set III (Pickford 1981). Nevertheless, the evolutionary level of some suids implies a later age for Gebel Zelten (Pickford 1987). Finally, if we try to correlate Gebel Zelten with European mammal zones, we can support a MN4 level. This locality has also yielded some rodents. Among them, the primitive murid Potwarmus sp. which looks quite similar to P. primitivus from Siwaliks of the Kamlial, northern Pakistan. P. primitivus has a time range between 14-3 and 18-0 Ma (Flynn et al. 1995). Almost the same conclusion is reached by Thomas (1979). de BONIS ET A L.\ MIOCENE GIRAFFID 131 The locality of Thymiana in Chios can be dated to the middle Miocene, and more precisely MN5, based on the evolutionary stage of its rodents. Indeed, the rodent faunas collected from two levels, a few metres below and above the large mammal locality, contain several elements unknown in MN4 localities, such as ctenodactylids, five species of cricetids, the genus Alloptox (Ochotonidae, Lagomorpha). On the other hand, the stage of evolution of cricetids and glirids from Chios is much less progressive than those from MN6 localities. The second argument for the age of the Chios faunas has been provided by magnetic stratigraphy. The mammalian localities are all included in a long reverse zone (Kondopoulou et al. 1993) which is preceded by an alternation of short normal and reverse polarities. The long normal zone including the fossiliferous levels is correlated tentatively with the Chron C5Br. According to Cande and Kent ( 1 995), the time span of this chron is between 1516 and 1 6 03 M a. Such an age is in good agreement with the correlation suggested for the MN5 zone by Steininger et al. (1990, 1996). Whatever the dating of the fossiliferous localities, both species G. georgalasi and C. syrtensis together with both species of Injanatherium and with Giraffokeryx constitute a morphological series, which may be a phyletic one, and which can be formalized as the subfamily Canthumericinae. Acknowledgements . The Chios project has been financed by the French CNRS, the National Research Centre of Greece and the Aristotle University of Thessaloniki. We thank for access to specimens from Gebel Zelten in The National History Museum, Dr J. J. Hooker and Dr P. Andrews; especially Dr A. W. Gentry for useful comments on the manuscript ; Dr H. Thomas for showing one of us the original Injanatherium hazimi material in the Museum national d'Histoire naturelle, Paris. We thank also Dr Andrew R. Milner and two anonymous reviewers whose remarks and criticisms greatly improved the manuscript. We give many thanks to Sabine Riffaut for the drawings and to Ghislaine Florent for the preparation of the manuscript. REFERENCES aguirre, e. and leakey, p. 1974. Nakali, nueva fauna de Hipparion del Rift Valley de Kenya. Estudios Geologicos , 20, 219-227. arambourg, c. and magnier, p. 1961. Gisements de vertebres dans le bassin tertiaire de Syrte (Lybie). Comptes Rendus des Seances de FAcademie des Sciences , 252, 1 181-1 183. astibia, p. h. and morales, j. 1987. Triceromeryx turiasonensis nov. sp. (Palaeomerycidae, Artiodactyla, Mammalia) del Aragoniense medio de la cuenca del Ebro (Espana). Paleontologia i Evolucio , 21. 75-115. bohlin, b. 1926. Die Familie Giraffidae. Paleontologia Sinica , C, 4, 1-178. cande, s. c. and kent, d. v. 1995. Revised calibration of the geomagnetic polarity timescale for the late Cretaceous and Cenozoic. Journal of Geophysical Research, 100, B4, 6093-6095. churcher, c. s. 1970. Two new upper Miocene giraffids from Fort Ternan, Kenya, East Africa. 1-105. In leakey, L. s. b. and savage, R. J. G. (eds). Fossil vertebrates of Africa. Academic Press, London, 333 pp. 1978. Chapter 25. Giraffidae. 509-535. In maglio, v. j. and cooke, h. b. s. (eds). Evolution of African mammals. Harvard University Press, 641 pp. ciric, a. and thenius, e. 1959. Uber das Vorkommen von Giraffokeryx (Giraffidae) im europaischen Miozan. Anzeiger der Akademie Wissenschaften mathematisch und naturwissenschaftliche Klasse, Wien , 9, 153-162. Colbert, e. h. 1933. A skull and mandible of Giraffokervx punjabiensis Pilgrim, American Museum Novitates, 632, 1-14. — 1936. Palaeotragus in the Tung Gur Formation of Mongolia. American Museum Novitates, 874, 1-17. crusafont-pairo, m. 1952. Los jirafidos fosiles de Espana. Memorias y Comunicationes de! Instituto Geologico, 8, 1-239. FLYNN, L. J., BARRY, J. C., MORGAN, M. E., PILBEAM, D., JACOBS, L. L. and LINDSAY, E. H. 1995. Neogene Siwalik mammalian lineages, species longevities, rates of change, and modes of speciation. Palaeogeography, Palaeoclimatology, Palaeoecology, 115, 249-264. gentry, a. w. 1990. Ruminant artiodactyls of Pasalar, Turkey. Journal of Human Evolution, 19, 529-550. — 1994. The Miocene differentiation of old world Pecora (Mammalia). Historical Biology, 7, 1 15-158. geraads, d. 1986. Remarques sur la systematique et la phylogenie des Giraffidae (Artiodactyla, Mammalia). Geobios , 19, 465^477. 132 PALAEONTOLOGY, VOLUME 40 godina, a. y., vislobokova, i. a. and abdrachmanova, l. t. 1993. A new representative of the Giraffidae from the Lower Miocene of Kazakhstan. Palaontologicheski Zhurnal , 1993, 75-86. Hamilton, w. r. 1973. The Lower Miocene ruminants of Gebel Zelten, Libya. Bulletin of the British Museum (Natural History ), Geology Series, 21, 76-150. - 1978. Chapter 24. Cervidae and Palaeomerycidae. 496-508. In maglio, v. j. and cooke, h. b. s. (eds). Evolution of African mammals. Harvard University Press, 641 pp. heintz, e. 1976. Les Giraffidae (Artiodactyla, Mammalia) du Miocene de Beni Mellal, Maroc. Geologie Mediterraneenne, 3, 91-104. brunet, m. and sen, s. 1981. Un nouveau giraffide du Miocene superieur d’lrak, Injanatherium hazimi n.g., n. sp. Comptes Rendus des Seances de I Academic des Sciences, 292, 423 — 426. kondopoulou, d., bonis, l. de, koufos, g. d. and sen, s. 1993. Palaeomagnetic data and biostratigraphy of the middle Miocene vertebrate locality of Thymiana (Chios Island, Greece). Proceedings of the Second Congress of the Geophysical Society of Greece, 2, 626-635. koufos, g. d., bonis, l. de and sen, s. 1995. A new viverrid (Carnivora, Mammalia) from the middle Miocene of Chios Island, Greece. Geobios, 28, 511-523. lehmann, u. and tobien, h. in press. Artiodactyle fossilien (Mammalia) aus dem Miozan von Thymiana, Chios. Annales Geologiques des pays Helleniques. melentis, J. and tobien, h. 1967. Palaontologische Ausgrabungen auf der Insel Chios (eine vorlauhge Mitteilung). Proceedings of the Academy of Athens, 42, 147-152. morales,, J., soria, d. and thomas, h. 1987. Les Giraffidae (Artiodactyla, Mammalia) d'Al Jadidah du Miocene moyen de la formation Hofuf (Province du Hasa, Arabie Saoudite). Geobios, 20, 441^467. moya-sola, s. 1986. El genero Hispanomeryx Morales et al. (1981), posicion filogenetica y systematica. Su contribucion al conocimiento de la evolucion de los Pecora (Artiodactyla, mammalia). Paleontologia i Evolucio, 20, 267-287. 1987. Los ruminantes (Cervoidea y Bovoidea, Artiodactyla, Mammalia) del Ageniense (Mioceno inferior) de Navarrete del Rio (Teruel, Espana). Paleontologia i Evolucio, 21, 247-269. paraskevaidis, i. 1940. Eine obermiocane Fauna von Chios. Neues Jahrbuch fur Mineralogie, Geologie und Paldontologie, Abteilung B, 83, 363-442. pavlovic, m. b. 1969. Miozan-Saugetiere des Toplica-Beckens. Annales Geologiques de la Peninsule des Balkans, 34, 269-394. [In Serbo-Croat, German summary], pickford, M. 1981. Miocene mammalian biostratigraphy for Western Kenya. Journal of Human Evolution, 10, 73-97. - 1987. A revision of the Miocene Suidae and Tayassuidae (Artiodactyla, Mammalia) of Africa. Tertiary Research, Special Paper, 7, 5-83. pilgrim, G. 1910. Notices on new mammalian genera and species from the Tertiary of India. Record of the Geological Survey, 40, 63-71. - 1911. The fossil Giraffidae of India. Memoirs of the Geological Survey of India, Paleontologia Indica, New Series, 4, 1-29. 1934. Correlation of ossiferous sections in the upper Cenozoic of India. American Museum Novitates, 704, 1-5. qiu zhanxiang, yan defa, jia hang and sun bo 1985. Preliminary observations on the newly found skeletons of Palaeomeryx from Shanwang, Shandong. Vertebrata Palasiatica , 23, 173-195. [In Chinese, English summary]. savage, r. 1989. The African dimension in European early Miocene mammals faunas. In lindsay, e. h., fahlbusch, v. and mein, p. (eds). European Neogene mammal chronology. NATO ASI ser.. A, Life Sciences, 180, 587-599. selley, r. 1969. Near-shore marine and continental sediments of the Syrte basin, Libya. Quarterly Journal of the Geological Society, London, 124, 1 49 — 460. steininger, f. f., bernor, r. l. and fahlbusch, v. 1990. European Neogene marine/continental chronologic correlations. In lindsay, e. h., fahlbusch, v. and mein, p. (eds). European Neogene mammal chronology. NATO ASI ser. A, Life Sciences, 180, 15^46. - berggren, w. b., Kent, d. v., bernor, R. l., sen, s. and agusti, J. 1996. Circum-mediterranean Neogene marine/continental chronologic correlations of European mammals. 7^46. In bernor, r. l., fahlbusch, u. and mittman, w. (eds). Evolution of Neogene continental biotopes in Europe. Columbia University Press, 746 pp. thomas, h. 1979. Le role de barriere ecologique de la ceinture saharo-arabique au Miocene: arguments paleontologiques. Bulletin du Museum national d Histoire Naturelle , 4e serie, 1, C, 2, 127-135. de BONIS ET AL.\ MIOCENE GIRAFFID 133 tobien, h. 1968. Palaontologische Ausgrabungen nach jungtertiares Wirbeltieren auf der Insel Chios (Griechenland) und bei Maragheh (NW-Iran). ‘ Jahrbuch Verenieing Freunde der Universitdt Main:', 51-58. 1980. A note on the skull and mandible of a new choerolophodont mastodon (Proboscidea, Mammalia) from the middle Miocene of Chios (Aegean Sea, Greece). 299-307. In Jacobs, l. j. (ed.). Aspects of vertebrate history. Essays in honor of Edwin Harris Colbert. Flagstaff, Arizona, 407 pp. LOUIS de BONIS Universite de Poitiers Laboratoire de Geobiologie, Biochronologie et Paleontologie Humaine 40 avenue du Recteur Pineau 86022 Poitiers Cedex, France GEORGE D. KOUFOS Aristotle University of Thessaloniki Department of Geology and Physical Geography 540 06 Thessaloniki, Greece SEVKET SEN Institut de Paleontologie Museum National d’Histoire Naturelle Typescript received 5 February 1996 8 rue de ButTon Revised typescript received 8 July 1996 75005 Paris, France A NEW PLIOSAUR FROM THE BAJOCIAN OF THE NEUQUEN BASIN, ARGENTINA by ZULMA GASPARINI Abstract. A new pliosaur with a spatulate symphysis is described as Maresaurus cocccii gen. et sp. nov. The holotype is from the upper levels of the Los Molles Formation (Lower Bajocian, Middle Jurassic) of the southern Neuquen Basin, central-western Argentina. Maresaurus coccai shares with Simolestes such synapomorphies as the absence of a dorso-medial foramen, lack of anterior interpterygoid vacuity, and fewer than 26 alveoli in the dentary. The autapomorphies of M. coccai are the strong flanges formed by the premaxillary and maxillary, the expanded and elevated pterygoid wings, a diastema between the eighth and ninth mandibular alveoli and the hypertrophied teeth (caniniforms) which are densely striated and without carinae or smooth faces. The record of Bajocian plesiosaurs is restricted to Simolestes keileni from the Upper Bajocian of France and M. coccai from the Lower Bajocian of the Neuquen Basin. The latter is associated with other marine reptiles, mainly ichthyosaurs, coeval with greater diversity in the invertebrate fauna, in an offshore deposit dominated by pelagic sedimentation. Most Jurassic Pliosauroidea ( sensu Brown 1981) have been found in Europe, mainly in the Lias of England and Germany (Fraas 1910; Taylor 1992a, 19926), in the Lower and Middle Oxford Clay (Middle-Upper Callovian) of England (Andrews 1913;Tarlo 1960; Brown 1 98 1 ; Martill 1991) and in the Oxfordian and Kimmeridgian of England and France (Taylor and Benton 1986; Bardet 1992, 1993; Mazin el al. in press). Outside the European Tethys region, the record of Jurassic pliosauroids is poor (Persson 1963 ; Gasparini 1985, 1992; Bardet 1992; Gasparini and Fernandez 1996, in press). However, after the Lias and before the Callovian, pliosaur records are particularly sparse. Hitherto, only a pliosaurid from the Upper Bajocian of Lorraine, France ( Simolestes keileni Godefroit, 1994), and a supposed rhomaleosaurid (sensu Kuhn 1961) from the Middle Jurassic of the Sichuan Basin, China ( Yuzhoupliosaurus chengjiangensis Zhang, 1985) have been published. Here, I describe a new pliosaurid from the Neuquen Basin. Pliosauridae (sensu Brown 1981) with spatulate snouts are known from: the Hettangian of England, Rhomaleosaurus megacephalus (Stutchbury, 1846) (see Cruickshank 1994), and Eurycleidus arcuatus ( Owen 1840); the Lower Toarctan of England, Rhomaleosaurus thorntoni (Andrews, 1922), R. zetlandicus (Phillips in Anon. 1854, fide Taylor 1992a, 19926), R. propinquus (Watson, 1910) and R. cramptoni (Carte and Baily, 1863); and the Upper Toarcian of Holzmaden, Rhomaleosaurus victor (Fraas, 1910). Recently, Cruickshank (1994) noted that R. zetlandicus , R. cramptoni and R. thorntoni could represent a single species. Part of a postcranial skeleton found in the Upper Lias of the Sichuan Basin (Bishanopliosaurus youngi Dong, 1980) was also referred to a pliosaurid (Rhomaleosauridae sensu Kuhn 1961). Other pliosaurids with a spatulate symphysis are: Simolestes keileni , from the Upper Bajocian of France (Godefroit 1994); Simolestes vorax Andrews, 1909, from the Middle Callovian of England (Martill 1991) and from the Lower Callovian of Normandy (Bardet 1993) and the Calcaires Blancs du Poitou (Mazin et al. in press); Simolestes sp. from the Callovian of the Moscow Basin (Tarlo 1960); Yuzhoupliosaurus chengjiangensis Zhang, 1985 from the Middle Jurassic of the Sichuan Basin; and Simolestes indicus (Lydekker) (Bardet et al. 1991) from the Tithonian of Kachchh, India. Simolestes nowackianus von Huene, 1938, from the Oxfordian of Ethiopia, has been reinterpreted as a specimen of the teleosaurid crocodile Machimosaurus (Bardet and Hua in press). Bardet (1992), Taylor (19926) and Godefroit (1994) have pointed out that the taxonomy of Rhomaleosaurus and other pliosauroids needs to be (Palaeontology, Vol. 40, Part 1, 1997, pp. 135-147] © The Palaeontological Association 136 PALAEONTOLOGY, VOLUME 40 reviewed. There were pliosaurids with a spatulate symphysis throughout the Jurassic across a wide sector of the Tethyan belt, but the new pliosaurid from the Lower Bajocian of the Neuquen Basin is the first record from the Eastern Pacific. The Early Jurassic pliosauroids with a spatulate symphysis were included within the family Rhomaleosauridae Kuhn, 1961 (Persson 1963), an invalid taxon (Brown 1981) because of the absence of unequivocal synapomorphies supporting it. Early in 1988, the author undertook fieldwork in the area of Chacaico Sur (Sierra de Chacaico), Neuquen province, north-western Patagonia, accompanied by Dr Luis Spalletti, sedimentologist of the Instituto de Investigaciones Geologicas from the Universidad Nacional de La Plata, Messrs Sergio and Rafael Cocca, technicians of Museo Profesor Olsacher of Zapala, Neuquen, and local people. That fieldwork covered the upper section of the Los Molles Formation (Cuyo Group; Dellape et al. 1978), referred to the Lower Bajocian (Spalletti et al. 1994). The Neuquen Basin is on the western margin of South America (Eastern Pacific), (32—41° S, 68-72° W). The top of the Los Molles Formation in the Chaciaco Sur area is composed of dark shales and marls with sandstone intercalations. This section, where the new pliosaurid was discovered, is interpreted as an offshore deposit dominated by pelagic sedimentation, with the sands, introduced by storm-induced orbital and gravitational flows (Spalletti et al. 1994). SYSTEMATIC PALAEONTOLOGY Superorder sauropterygia Owen, 1860 Order plesiosauria de Blainville, 1835 Superfamily pliosauroidea (Seeley, 1874) Welles, 1943 Family pliosauridae Seeley, 1874 Genus maresaurus gen. nov. Derivation of name. From mare, the Latin word for sea, and sauros , the Greek word for lizard. Type species. Maresaurus coccai sp. nov. Diagnosis. As for the type and only species. Maresaurus coccai gen. et sp. nov. Text-figures \-b 1993 Simolestes sp. Gasparini and Fernandez, p. 107 Derivation of name. Dedicated to the brothers Sergio and Rafael Cocca, members of the Museo Prof. Olsacher of Zapala, and valuable collaborators in all the fieldwork related to the search for marine reptiles in the Neuquen Basin. Holotype. Museo Prof. Olsacher, Zapala, Neuquen (MOZ 4386 V), articulated skull and mandible, fused atlas and axis, and the first cervical vertebrae. Locality and horizon. Chacaico Sur, (39° 15' S, 70° 18' W), 70 km south-west of Zapala, Neuquen Province, Argentina (Spalletti et al. 1994); upper part of the Los Molles Formation, Cuyo Group (Dellape et al. 1978; Leanza 1990; Riccardi and Gulisano 1990). The new pliosaurid comes from a level of dark shales and marls. Within the Emileia giebeli , Emileia multiformis subzone. Lower Bajocian, Middle Jurassic (Spalletti et al. 1994). Diagnosis. Pliosaurid with spatulate symphysis incorporating six pairs of alveoli. Rostrum deep, with marked sagittal crest, formed by union of premaxillae, and two conspicuous parallel crests, formed by dorsal union of premaxillae and maxillae. Deep notch between premaxilla and maxilla, and marked anterior wave in maxilla incorporating six alveoli. External surface of the premaxillae and maxillae with deep cavities. No dorsomedial foramen in premaxillae. Posterior region of the GASPARINI: BAJOCIAN PLIOSAUR 137 text-fig. 1. Maresaurus coccai gen. et sp. nov.; MOZ 4386 V; Lower Bajocian, Neuquen Basin, Argentina. A, lateral view; b, dorsal view; c, ventral view. Scale bar represents 100 mm. 138 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Maresaurus cocccii gen. et sp. nov.; MOZ 4386 V; Lower Bajocian, Neuquen Basin, Argentina. Scale bar represents 200 mm. For abbreviations see Appendix. GASPARINI: BAJOCIAN PLIOSAUR 139 parietal wide, without enclosing posterior edge of skull and not extending below dorsal ramus of squamosal. Occipital condyle not exposed in dorsal view. No anterior interpterygoid vacuity. Very high pterygoid posterior wings. Parasphenoid not separating interpterygoid vacuities completely. Twenty-four teeth in dentary. Anisodonty. All the teeth with circular section, densely distributed non-dichotomized striae, and without carinae. Hypertrophied (caniniform of Tarlo 1960) teeth without smooth faces. DESCRIPTION Skull and mandible The skull is sub-triangular in dorsal view, with a rather short and spatulate rostrum as in Rhomaleosaurus and Simolestes. In lateral view, the rostral height, the cranial height (which reaches its maximum in the squamosals), the notch between premaxillae and maxillae, and the strong maxillar wave are important features (Text-fig. 1). The medial sector between the external nares and the squamosals is damaged by erosion and compressional effects after fossilization, as the parietal crest was the only bony element exposed on the surface. Most dorsal cranial sutures are missing, and the temporal arches are lost. In contrast, the palate and the mandible are excellently preserved. The mandible is strongly attached to the skull and rostrum, in part due to the large, partly intermeshed teeth. Premaxillae. The premaxillae, the lateral edges of which are anteriorly slightly convex, form the anterior part of the snout which is blunt-ended. In this anterior region, there are several hollows irregularly distributed and pores near the dentary margin. In the posterior part of the rostrum and above the maxillae, the premaxillae form a crest which, at the anterior edge of the external nares, bifurcates into two gentle domes which merge with the frontals or parietals. Unlike Rhomaleosaurus (Carte and Daily 1863; Taylor 19926; Cruickshank 1994) and Simolestes (Andrews 1913), the contact between premaxillae and maxillae produces a strong crest which runs from the notch up to the anterior edge of the naris. Between the premaxillary-maxillary contact crests and the medial premaxillary crest there are two deep depressions, ornamented with irregularly distributed hollows in the anterior part, and tiny striae running anteroposteriorly (Text-fig. 1b). Unlike Rhomaleosaurus , Maresaurus coccai lacks a dorso-medial foramen above the premaxillae. Ventrally, the premaxillae are covered by the mandibular symphysis which is wider than the anterior edge of the snout. However, the alveoli of the functional teeth, five in each premaxilla, are exposed. The amount of dorsal extension of the premaxillae and their degree of participation in the narial margins, are unknown. As the medial crest of the premaxillae is damaged at the anterior edge of the external nares, the premaxillary domes are slightly depressed, and the frontal and prefrontal sector is partially eroded, it is impossible to ascertain whether the premaxillae were dorsally in contact with the frontals, or directly with the parietal (Text-fig. 1b). If the dorsal contact was with the parietal, the frontals would have remained laterally situated, as in Rhomaleosaurus zetlandicus. Unfortunately, details of this region are obscure in both R. cramptoni and in Simolestes vorax. In the latter, Andrews (1909, p. 427) pointed out that the premaxilla and frontal merge by the anterior edge of the orbits, but he later stated (Andrews 1913, p. 26) that the contact was made between the premaxillae and the parietal, with the frontals left outside. Following examination of the holotype (BMNH R3319), I concur with the latter opinion. In Maresaurus coccai , the frontals form the upper margin of the orbits and, in lateral view, they have deep furrows and crests where the postorbital arches, which are missing, were attached (Text-fig. 1a). Maxillae. The maxillae are almost complete, especially the right one, the alveolar edge of which is preserved up to mid-orbit level. The lateral walls of the maxillae are high, and bear, as in many crocodiles, a marked anterior wave which comprises the first six alveoli (Text-figs 1a, 2a). In Rhomaleosaurus thorntoni and Simolestes vorax , this maxillary wave is less evident, probably as a result of flattening. The large oblong cavities which cross the maxillae from posterodorsally to anteroventrally are a remarkable peculiarity. Similar pronounced cavities are also present in R. victor and R. cramptoni. The similarity of the sculpture of R. cramptoni to that of some crocodiles was pointed out by Carte and Baily (1863). The lateral contact between maxillae and premaxillae forms a strong crest starting with a marked marginal notch where a hypertrophied mandibular tooth was housed (the fifth or sixth). Each maxilla forms the lower edge of the external naris. A smooth channel commences at the anterior edge of the external naris, and extends forwards disappearing at the level of the fourth maxillary tooth. Ventrally, the mandibular rami almost cover the maxillae. However, it can be observed that the right maxilla starts at the external border of the internal naris. The exact number of alveoli is not known, because the left maxilla is preserved back to a point just anterior to the orbit, and the right one only half this distance (with 13 alveoli). Nevertheless, taking into account the extension of the 140 PALAEONTOLOGY, VOLUME 40 mandibular alveolar series, the maxillae had to extend up to the posterior part of the orbit, with approximately 22 alveoli for functional teeth. Nares. The external nares are anterior to the orbits, level with seventh and ninth maxillary alveoli (in R. thortoni, BMNH R4853, the external nares are situated further anteriorly, as the anterior edge is by the fifth maxillary alveoli). They are oval, with the anterior-posterior diameter longer. Preservational effects have made their sizes different, the right one being longer than the left one. The nares are bordered by the premaxillae at the antero-superior edge and by the maxillae at the antero-inferior one (Text-fig. 2b). The posterior edge in both nares cannot be determined because of the lack of sutures in this area, but dorsally they are limited by the frontals, and on the right naris, a small fragment might belong to the prefrontal. The presence of lacrimals in plesiosaurs has been a matter of debate. Taylor (19926, fig. 1) demarcated them with a dotted line in R. zetlandicus , while Storrs (1991) stated that the lacrimals were absent in most Sauropterygia, including plesiosaurs. However, lacrimals can be clearly observed in a new material of Pliosaurus brachyspondylus (Owen) (Taylor and Cruickshank 1993), or either may be fused to the maxilla as in Rhomaleosaurus megacephalus (Cruickshank 1994). In Maresaurus (MOZ 4386 V), there is a short suture line on the right side, which might coincide with the lacrimal anterior external edge. Furthermore, a channel which runs between the posterior external end of the naris and the anterior edge of the orbit, might belong to the lacrimal duct. Outside the naris, and over the maxilla (or lacrimal), there is a pair of large foramina (Text-figs 1b, 2b). No other known plesiosaur bears such foramina, so it is uncertain whether they are an autapomorphy of Maresaurus coccai , related to an osmoregulation mechanism (salt gland) or were produced by the fossilization process, that strangely left two almost symmetrical holes. Unfortunately, the lacrimal and prefrontal region in other pliosaurids with a spatulate symphysis is also either badly preserved or unknown. Furthermore, the problem of whether or not plesiosaurs possessed nasals is not resolved (Andrews 1913; Brown 1981 ; Storrs 1991 ; Taylor 19926; Brown and Cruickshank 1994). Parietal. If the parietal contacted the premaxillae, forming the interorbital bridge, it must have been narrow, ascending postero-dorsally after the contact. The pineal foramen, closed by a post-mortem break, would be placed in the parietal, just behind the postorbital arch. Although the back of the parietal is also eroded, a medial line can be observed, indicating that they were paired (at least internally, and probably visible in young specimens), in the narrower sector, in the anterior part of the bridge that divides the temporal fenestra. There is no interdigitated medial suture as in R. zetlandicus (Taylor 19926, p. 251). Here, in lateral view, the parietal has an irregular structure with furrows and small anterior-posterior crests which serve as a union with the epipterygoid (Text-fig. 2a). Postero-dorsally, the parietal widens abruptly, it is markedly convex and ascends to the squamosals. A medial bump marks the end of what was a gentle crest of the parietal (Text-fig. 1b). As in Rhomaleosaurus , the parietal neither forms a narrow sagittal crest, nor reaches the posterior edge of the skull (Text-tigs 1b, 2b). This does occur in Simolestes where a thin parietal bar is enclosed by the squamosals in the occipital edge of the skull. The posterior part of the parietal is covered by the squamosals so it can be observed in ventral and occipital view. It is a wide sub-triangular plate the vertex of which coincides with the sagittal line and is forwardly oriented. The ventral parietal plate has a medial crest and a pair of lateral shorter crests (Text- fig- 3). Squamosals. The squamosals merge in the dorsal cranial roof, fusing with the parietals with a wide interdigitated suture. They form the highest part of the skull. The parietals do not extend under the anterodorsal crest of the squamosals as in R. zetlandicus (Taylor 19926). Laterally, each squamosal forms the posterior edge of the temporal fenestra (the right fenestra is incomplete; Text-fig. 2a). This ramus is stout as in Rhomaleosaurus , while in Simolestes it is wider and more compressed. The dorsal rami are broken and the ventral ones are quite robust and are fused to the quadrates. Both dorsal surfaces are eroded and consequently the squamosal-quadrate suture is unclear; nor have the foramina been preserved, as in the R. zetlandicus holotype (Taylor 19926). The ventral ramus is proportionately narrower than in Simolestes. As in R. zetlandicus and some plesiosauroids (Brown 1981 ; Taylor 19926) the distal ramus of the paroccipital process is fixed to the medial face of the squamosal ventral ramus up to the quadrate suture. Quadrate. The quadrate is massive and fused to the mandibular glenoid fossa (Text-fig. 3). It is covered dorsally by the descendent or ventral ramus of the squamosal, and laterally, in posterior and internal view, it surrounds the outer extremity of the paroccipital process. Internally, the quadrate covers dorsally the distal portion of the quadrate ramus of pterygoid. In that medial and posterior sector of the left quadrate there is a marked boss. On the right side, this boss is missing and represented by a hole. The quadrate condyles are partially broken GASPARINI: BAJOCIAN PLIOSAUR 141 text-fig. 3. Maresaurus coccai gen. et sp. nov.; MOZ 4386 V; Lower Bajocian, Neuquen Basin, Argentina; occipital view. Scale bar represents 200 mm. For abbreviations see Appendix. and as they are fused to the glenoid fossa, their morphology is not clear. They are wide but not very marked and both of them appear to end in slightly concave surfaces. Epipterygoid. The epipterygoid has a wide base and lies dorsally on the parietal. Then it narrows into a stem or columella that runs downwards and fuses with the pterygoid anterior bar (Text-figs 1a, 2a). The left epipterygoid is somewhat displaced, but better preserved, and the quadrate ramus of pterygoid is fused to its posterior edge. Palate. The palate of Maresaurus is closed (Text-figs lc, 2c). Unlike those of R. victor and R. zetlandicus , it has no anterior interpalatal vacuity (Fraas 1910; Taylor 1992h). The tear-drop-shaped internal nares, are separated by the vomers, which are fused forming a convex bar. This bar continues forwards between the maxillae, which are almost covered by the mandibular rami. Behind the internal nares, the vomers widen and maintain a suture in the sagittal line, suggesting that they were paired, at least in the early ontogenetic developmental stages. Posteriorly the vomers make contact with the maxillae, palatines and pterygoids. The nares are bordered by vomers and maxillae; the palatines, as in R. zetlandicus, R. victor and Simolestes vorax, are not included. The flattened palatines merge outside the anterior rami of the pterygoids, reaching the anterior edge of a small suborbital fenestra elongated anterior-posteriorly and formed also by the external lateral and partially ventral ectopterygoid and the lateral ramus of the pterygoid. Both fenestrae are incomplete. The anterior edge of the subtemporal fenestra is beneath the limit formed by the ectopterygoid and the right pterygoid. Only the right ectopterygoid was preserved, partially superposed to the external margin of the palatine (Text-fig. 2c). Pterygoids. The pterygoids are complete and run between the palatines, up to the seventh pair of maxillary alveoli, where they unite with the vomer and palatines with an interdigitated suture. The oblong interpterygoid vacuities, with the anterior-posterior diameter much longer, are in the posterior third of the pterygoid anterior rami. These vacuities have a slight separation in the posterior region, as in R. zetlandicus, and are partially divided by a projection of the parasphenoid. In Simolestes, the parasphenoid crest does not reach the posterior region of the vacuities (Andrews 1913, pi. 3), while in Rhomaleosaurus the parasphenoid is a full crested bar which separates both fossae. The parasphenoid runs in front of the interpterygoid vacuities and between the pterygoids, as in Simolestes (Andrews 1913), R. megacephalus (Cruickshank 1994) and probably in R. zetlandicus (Taylor 1992 b). Laterally, and in front of the interpterygoid vacuities, the pterygoids expand and fuse with the ectopterygoids to form the posterior edge of the small vacuities mentioned above. The posterior part of the pterygoids is deeply concave due to the rise of the pterygoid boss (high posterior pterygoid wings). The pterygoid concavity is bigger and posteriorly more extended than in R. victor. In Simolestes vorax (holotype BMNH R3319) the pterygoid is flat. The pterygoids and their lateral long bars extend at both sides and below the palatine; posteriorly, they fuse to the anteromedial corner of the quadrate below the squamosal- quadrate contact. In the middle of their run they cover, in ventral view, the paroccipital processes. 142 PALAEONTOLOGY, VOLUME 40 Occipital region. The occipital region is incomplete (Text-fig. 3). The basioccipital forms the occipital condyle, which is slightly wider than high as in R. zetlandicus (Taylor 19926), but shorter. As in Rhomaleosaurus , the condyle cannot be seen in dorsal aspect because it is covered by the squamosals. In Simolestes it is exposed. In the laterodorsal region of the condyle, on both sides of the neural canal, there are two depressed surfaces for the lost exoccipitals. On both sides of the condyle, the basioccipital has descending expansions (lateral processes of basioccipital of Andrews 1913, or basipterygoid process of Brown 1981) which leant on both hollows of the pterygoid posterior rami. The paraoccipital processes, both incomplete, are flat and rest on the pterygoid posterior rami, ending level with the squamosal-quadrate suture. On the left side, in front of the basioccipital, there are two bony elements. The ventral one may correspond to the basisphenoid (see Kimmerosaurus langhami Brown, 1981). The supraoccipital is not preserved. Mandible. Both mandibular rami are preserved firmly attached to the skull and snout, and therefore the alveolar sector is partially covered by the premaxillae and maxillae. The mandibular symphysis is short and includes, in ventral view, six pairs of functional alveoli (it is important to define whether the number should be counted either from the ventral or the buccal view, since it can be different). The Maresaurus symphysis is wide (index: 097, considering the relationships between the symphysis maximum width over maximum length in ventral view), similar to that in R. zetlandicus (index: 1), but less than Rhomaleosaurus victor (index: IT 7), Simolestes keileni (index: IT 5) and S. indicus (index: IT 5) (Bardet et ah 1991 ; Godefroit 1994), and slightly wider than S. vorax (index: 086). The greater broadening of the symphysis in Maresaurus coccai is between the fifth and sixth alveoli, where it also rises in a similar degree to that of R. cramptoni (Carte and Baily 1863), greater than that of R. zetlandicus and less than in R. victor and the holotype of Simolestes vorax (in BMNH R3170 it is dorsally flattened). The snout does not cover the mandibular symphysis as in R. victor. Instead, in R. cramptoni , according to Carte and Baily (1863 pi. 7, fig. 1), it covers the symphysis. The mandibular sym- physis of Maresaurus narrows forwards and ends rounded, with only one pair of small alveoli; behind the sixth pair of alveoli the symphysis narrows abruptly and then separates into both mandibular rami. The splenials can be observed in ventral view, being part of the symphysis, level with the fourth pair of alveoli. The lateral and ventral walls of the symphysis have deep anterolaterally orientated holes and numerous foramina, those parallel to the alveolar edge coinciding with each interalveolar septum. Between the seventh and tenth alveoli, the dentary narrows where the vertical walls are the lower ones. Then, it separates and deepens until it reaches the coronoid eminence. The dentary has 24 alveoli. In medial view it is fused to the splenial and ventrally the angular reaches it at the tenth alveolus level. Posteriorly, it contacts the surangular, although the suture is not preserved (Text-fig. 2a). The coronoid is exposed on the inner face of the mandible; the coronoid eminence is not conspicuous and appears behind the most posterior alveoli, while the maxilla covers it externally. In medial view, the coronoid contacts anteriorly with the splenial and posteriorly with the surangular above, and the prearticular below. The prearticular is long, since it extends from the intermandibular foramen probably up to the articular, but there are no sutures that prove whether the prearticular was part of the glenoid fossae anterior wall. The prearticular has been identified both in pliosauroids (R. zetlandicus ; Taylor 19926) and in plesiosaurids ( AIzdasaurus colombianus Welles, 1962). The angular is another long element that extends ventrally and laterally to the tenth alveolus. Externally, it forms the expanded base of the adductor fossa, incised mainly in the surangular, and serves as support for the articular (glenoid fossa). In front of the glenoid zone there are no sutures demarcating the bones. Consequently it is impossible to know whether this region was bordered by part of the surangular and the prearticular or only by the articular, and the extent of participation, if any, of the angular in the retroarticular process. The retroarticular processes are dorsally trapezoidal, with flat surface (Text-figs 1b, 3). However, their bodies are high and ventrally convex (Text-fig. 1a). These retroarticular processes are like those of R. zetlandicus , but differ from those of R. victor which are compressed and high, and are very different from those of Simolestes vorax which are flat. Though the retroarticular processes of S. keileni are damaged, their proportions are similar to those of Maresaurus and different from those of Simolestes vorax. Dentition Each premaxilla bears five alveoli for functional teeth. The first is small and medial, the second is twice as large, and the third, fourth and fifth held hypertrophied teeth (caniniform sensu Tarlo 1960). Posteriorly, there is a large interalveolar space which corresponds to the premaxilla-maxilla union. The hypertrophied sixth mandibular tooth fitted into this space. The first alveolus of the maxilla is small, the second is slightly larger, the third, fourth and fifth are hypertrophied, especially the fourth, which coincides with the maxillary wave GASPARINI: BAJOCIAN PLIOSAUR 143 maximum inflexion. From the sixth alveolus posteriorly, they decrease in size. Both maxillary rami are incomplete; the right one bears 19 alveoli and the left one 14. Flowever, considering the jaw dental series, the maxillary teeth may have numbered 24. The dentaries retain the entire alveolar line (25 in the left side and 24 + on the right side) and several tooth remains in different developmental stages. The mandibular symphysis bears in ventral view, six pairs of alveoli, the first of which is small, the second and third, bigger, and the fourth to sixth, hypertrophied. From the seventh pair of alveoli, there is a narrowing with reduction of the alveoli, coincident with the separation of the mandibular rami. Posterior to the eighth pair of alveoli, there is a wide interalveolar space, visible in the left dentary, where the third maxillary hypertrophied tooth was housed. This is not seen in Rliomaleosaurus , where there are no spaces or diastemas. Finally, the series ends with 13 medium- sized alveoli of which the last three are the smallest. All the alveoli are circular to sub-circular, in accordance with the cross section of the functional teeth. In the rear of the left mandibular ramus, several small alveoli are preserved which correspond to replacement teeth. As in all plesiosaurs, these replacement teeth are placed medially and posteriorly to the alveoli of the functional teeth. Based on the opening or occlusion degree of these small alveoli, there appears to have been alternate tooth replacement in waves, from back to front (Edmund 1960). Several teeth were detached during the excavation. The most complete is a hypertrophied one (Text-fig. 4a), circular in cross section, backwardly recurved, with an incomplete root without carinae, striated crown and the apex with only four striae. The striae are not dichotomously branching; they have a compact distribution (20 striae per 10 mm in the middle of the crown), and the pattern is of one (occasionally two to three) shorter stria between longer ones, of which only four reach the smooth apex. One pair of long striae is situated on the lateral face of the crown and the other pair on the inner face. In spite of being hypertrophied, this tooth has no smooth faces as in Rhomaleosaurus zetlcmdicus , R. cramptoni, Simolestes keileni and S', vorax. As in all pre- Kimmeridgian pliosaurids, all the teeth cross sections of Maresawus are circular. The Kimmeridgian- Tithonian specimens have triangular sections (Tarlo 1960). Vertebrae The atlas, axis and three articulated anterior cervical vertebrae are the only postcranial skeletal elements preserved (Text-fig. 4b), together with two other cervical vertebrae, also from the anterior sector of the neck. The atlas and axis are fused by their lateral faces, but not ventrally, where they are separated by a deep furrow. I— 1 text-fig. 4. Maresaurus coccai gen. et sp. nov. ; MOZ 4386 V ; Lower Bajocian, Neuquen Basin, Argentina, a, caniniform tooth, b, atlas-axis and cervical vertebrae. Scale bar represents 20 mm (a), 30 mm (b). 144 PALAEONTOLOGY, VOLUME 40 The anterior part of the atlas body is deeply concave. Laterally the atlas body is longer than the axis and there is a ventrolateral furrow which forms a ‘lip’ surrounding the occipital condyle. The left side of the atlas-axis is more damaged, but the right side preserves part of a long post-zygapophysis which reaches the first cervical vertebra. The three following cervical vertebrae are short, wider than long. In the first one, the rib is detached and it can be observed that the hollow remaining, belonging to the synapophysis is not divided; so, at this level of the neck, the ribs were unicipital. The two isolated vertebrae, which also belong to the anterior sector of the neck, show a small narrowing in the synapophysis, suggesting that the ribs there were bicipital. Both Rhomaleosaurus thorntoni and R. cramptoni have cervical vertebrae with a divided synapophysis. In both cases, the cervicals were not the more anterior ones (Carte and Baily 1863; Andrews 1922). The neural spine is low and strong in the more anterior cervicals and higher in the following ones. The pre- and post-zygapophyses are wide and low. DISCUSSION Phylogenetic analysis of sauropterygians has advanced significantly in recent years (Rieppel 1989; Sues 1989; Storrs 1991). However, according to Taylor and Cruickshank (1993), the inter- relationships between sauropterygians and plesiosaurs need defining with more precision, and important progress depends on the study of more Jurassic plesiosaurs. Brown (1981) made a significant contribution, carrying out a detailed analysis of plesiosaur characters. He selected some synapomorphies used to discriminate families, and removed other characters as either plesiomorphic or subject to ontogenetic variation. However, it is difficult to distinguish synapomorphies in Jurassic pliosaurids working only with skulls, because of the possibility that similarities could either be due to close phylogenetic links or convergences in response to similar feeding habits. Furthermore, there have been few studies on Jurassic pliosaurids with relatively complete skulls. The classic studies of Andrews (1913, 1922) were followed by the recent and outstanding reviews of Taylor (1992a, 1992 b), Taylor and Cruickshank (1993) and Cruickshank (1994). Pliosaurids are considered to be the sister-group of plesiosauroids because of their cheek structure (Brown and Cruickshank 1994). All Liassic pliosaurids are mesorostral (rostrum length, from the premaxilla to the orbit anterior edge/skull length, from the premaxilla to the squamosal: < 0-50). The mesorostry is shared by Rhomaleosaurus , Maresaurus and Simolestes , while Peloneustes, Liopleurodon and some species of Pliosaurus share the long rostrum (index > 0-50) as a derived character. However, the spatulate symphysis shared by Rhomaleosaurus , Maresaurus and Simolestes could be a convergent character of animals with similar feeding habits. So, at present, there is no synapomorphy known to support the monophyly of the spatulate genera, and in this sense, in agreement with Brown (1981), Rhomaleosauridae is considered not to be a valid taxon. Maresaurus and Rhomaleosaurus share: the parietal not reaching the occipital edge of the cranial plate and not forming the sagittal crest between the squamosals (probably a primitive character which is maintained in Pliosaurus); the occipital condyle not visible in dorsal view; and the squamosal dorsal ramus with elliptical cross section. In Simolestes, the parietal is a crest that reaches the posterior edge of the skull, the occipital condyle is visible in dorsal view, and the squamosal dorsal ramus is compressed and sub-triangular in cross section. Maresaurus shares with Simolestes (and Liopleurodon) derived characters such as the lack of dorso-medial foramen, the lack of anterior interpterygoid vacuity (both present in Rhomaleosaurus and Pliosaurus) and fewer than 26 alveoli in the dentary (Liassic pliosaurids and some species of Pliosaurus have at least 30). Among Jurassic pliosaurids, Rhomaleosaurus is probably the most primitive, and Simolestes, the most derived. Maresaurus shares characters with both. However, Maresaurus has autapomorphies which justify its differentiation at generic and specific level: the strong flanges formed by the pmx-mx; the parasphenoid that does not separate completely the interpterygoid vacuities; the expanded and elevated pterygoid wings; the hypertrophied teeth densely striated without carines and smooth faces; and a diastem between the eighth and ninth mandibular alveoli. If those foramina placed externally to the external nares are not artefacts, this is another autapomorphy of Maresaurus. In order to understand the phylogeny of Maresaurus , it is necessary to approach the revision of all the Jurassic pliosaurids using the same taxonomic criteria. GASPARINI : BAJOCIAN PLIOSAUR 145 Records of Bajocian marine reptiles are scarce world-wide (Bardet 1992; Gasparini 1992; Gasparini and Fernandez 1996). Even scarcer are records of plesiosaurs, which are restricted to Simolestes keileni from the Upper Bajocian of France (Godefroit 1994) and Maresaurus coccai from the Lower Bajocian of the Neuquen Basin. The extinction of the rich Early Jurassic marine reptile fauna in the European Tethys may have been a response to the shallowing of the basins. When lagoons prevailed, thalattosuchians were almost exclusive in the pre-Callovian marine herpetofauna (Bardet 1992; Vignaud 1995; Mazin et al. in press). On the contrary, the assemblage of marine reptiles in the Lower Bajocian in the Neuquen Basin comprises pelagic and off-shore forms, namely pliosaurids, at least two taxa of large ichthyosaurs, and a large thalattosuchian (Fernandez 1994; Spalletti et al. 1994). Accordingly, the palaeoenvironment is interpreted as an off- shore deposit dominated by pelagic sedimentation (Spalletti et al. 1994); and in the Emileia giebeli zone, where the marine reptiles were discovered, bivalves, brachiopods, ammonoids and microfossils attained high diversity (Riccardi et al. 1994). Acknowledgements. The author thanks the following persons and institutions which enabled this work to be carried out: in fieldwork. Dr Luis Spalletti (Centro de Investigaciones Geologicas, Universidad Nacional de La Plata), and Messrs Sergio and Rafael Cocca (Museo Prof. Olsacher, Zapala); for preparing the material, Jose Laza, Javier Posik and Omar Molina (Dept Paleontologia Vertebrados, Museo de La Plata); for drawings, Maximiliano Lezcano; for translation, Cecilia Deschamps and Susana Bargo (Museo de La Plata); for access to material, J. Garate Zubillaga and S. Cocca (Museo Prof. Olsacher, Zapala), Dr Angela C. Milner (The Natural History Museum, London), Dr D. B. Norman and Mr M. Dorling (Sedgwick Museum, Cambridge), Mr A. Dawn (Peterborough Museum), Dr M. Norell (American Museum of Natural History, New York) and Dr France de Broin (Museum National d'Histoire Naturelle, Paris). Finally, I am grateful to Dr Nathalie Bardet for helpful discussion and assistance with literature, and for reading and commenting on the manuscript. This research was supported by the National Geographic Society (Grant 4265-90, 5178-94) and Consejo de Investigaciones Cientificas y Tecnicas, Argentina (PID 3006200/88). REFERENCES Andrews, c. F. 1909. On some new Plesiosauria from the Oxford Clay of Peterborough. Annals and Magazine of Natural History , 4, 4 1 8^429. — 1913. A descriptive catalogue of the marine reptiles of the Oxford Clay. British Museum (Natural History), London, 2, 1-206. — 1922. Description of a new plesiosaur from the Weald Clay of Berwick (Sussex). Quarterly Journal of the Geological Society , London , 78, 285-298. bardet, n. 1992. Evolution et extinction des reptiles marins au cours du Mesozoique. Memoire des Sciences de la Terre , Universite Pierre et Marie Curie , Paris, 9230, 1-212. — 1993. Pliosaurs and plesiosaurs from the middle Jurassic (Callovian) of Normandy. Revue de Paleobiologie, 7, 1-17. — mazin, j.-m., carieu, E., enay, r. and Krishna, J. 1991. Les Plesiosauria du Jurasique superieur de la province de Kachchh (Inde). Comptes Rendus de TAcademie des Sciences, 313, 1343-1347. — and hua, s. in press. Simolestes nowackianus Huene, 1938 from the Upper Jurassic of Ethiopia reinterpreted as a teleosaurid crocodile, not a pliosaur. Neues Jahrbuch fur Geologie und Palaontologie. blainville, h. de 1835. Description de quelques especes de reptiles de la Californie, precedee de l’analyse d'un systeme d’Erpetologie et d’Amphibiologie. Nouvelles Annales Museum d'Histoire Naturelle , Paris, 4, 233-296. brown, d. s. 1981. The English Upper Jurassic Plesiosauroidea (Reptilia) and a review of the phylogeny and classification of the Plesiosauria. Bulletin of the British Museum ( Natural History), Geology Series, 35, 253-347. — and cruickshank, A. r. i. 1994. The skull of the Callovian plesiosaur Cryptoclidus eurymerus, and the sauropterygian cheek. Palaeontology, 37, 941-953. carte, a. and baily, w. 1863. Description of a new species of Plesiosaurus , from the Lias, near Whitby, Yorkshire. Journal of the Royal Dublin Society, 4, 160-170. cruickshank, a. r. i. 1994. Cranial anatomy of the Lower Jurassic pliosaur Rhomaleosaurus megacephalus 146 PALAEONTOLOGY, VOLUME 40 (Stutchbury) (Reptilia: Plesiosauria). Philosophical Transactions of the Royal Society of London, Series B, 343, 247-260. dellape, d., mombru, c., pando, G., riccardi, a., uliana, m. and westtermann, G. 1978. Edad y correlation de la Formation Tabanos en Chacay Melehue y otras localidades de Neuquen y Mendoza. Obra Centenario Museo de La Plata, 5 Paleontologi'a, 81-105. dong, zhiming. 1980. A new plesiosaur from the Lias of Sichuan Basin. Vertebrata PalAsiatica, 18, 191-197. Edmund, a. G. 1960. Tooth replacement phenomena in the lower vertebrates. Contributions to Life Sciences Division of the Royal Ontario Museum, 52, 1-190. Fernandez, m. 1994. A new long-snouted ichthyosaur from the early Bajocian of the Neuquen Basin (Argentina). Ameghiniana, 31, 291-297. fraas, e. 1910. Plesiosaurier aus dem oberen Lias von Holzmaden. Palaeontographica , 57, 105-140. gasparini, z. 1985. Los reptiles marinos jurasicos de America del Sur. Ameghiniana, 22, 22-34. 1992. Marine reptiles of the Circum-Pacific region. 361-364. In westermann, g. (ed.). The Jurassic of the Circum- Pacific, world and regional geology, 3. Cambridge University Press. and Fernandez, M. 1993. Reptiles marinos bajocianos en la Cuenca Neuquina. Ameghiniana, 30, 107. - 1996. Biogeographic affinities of the Jurassic marine reptile fauna of South America. 443-450. In riccardi, a. (ed.). Advances in Jurassic research. Georesearch Forum 1-2. Trans-Tech Publications, Switzerland. in press. Tithonian marine reptiles of the eastern Pacific. In Callaway, j. and nicholls, e. (eds). Ancient marine reptiles. Academic Press. godefroit, p. 1994. Simolestes keileni sp. nov., un pliosaure (Plesiosauria, Reptilia) du Bajocien superieur de Lorraine (France). Bulletin des Academie et Societe Lorraines des Sciences, 33, 77-95. huene, F. von 1938. Ein Pliosauride in Abessinien Central. Mineralogie, Geologie und Palaontologie, Abteilung B, 370-376. kuhn, o. 1961. Die Familien der rezenten und fossilen Amphibien und Reptilien. Verlanshaus Meinsenbach, Bamberg, 79 pp. leanza, h. 1990. Estratigrafia del Paleozoico y Mesozoico anterior a los movimientos intermalmicos en la comarca del Cerro Chachil, provincia del Neuquen. Revista de la Asociacion Geologica Argentina, 45, 272-299. martill, d. 1991. 10. Marine reptiles. 226-243. In martill, d. and Hudson, j. (eds). Fossils of the Oxford Clay. The Palaeontological Association London, Field Guides to Fossils, 4, 286 pp. mazin j.-m., bardet, N., viGNAUD, p. and hua, s. in press. Marine reptiles from the Middle and Upper Jurassic of western Europe. Paleovertebrata. owen, r. 1840. Report on British fossil reptiles. Part I. Report of the British Association for the Advancement of Sciences. 1839, 43-126. 1860. On the orders of fossil and Recent Reptilia, and their distribution in time. Report of the British Association for the Advancement of Sciences , 29, 153-166. persson, p. 1963. A revision of the classification of the Plesiosauria, with a synopsis of the stratigraphical and geographical distribution of the group. Lunds Universitets Arsskrift, N. F. 2, 59, 1-60. riccardi, a., damborenea, s., mancenido, m. and ballent, s. 1994. Middle Jurassic Biostratigraphy of Argentina. Geobios, Memoire Speciale, 17, 423—430. — and gulisano, c. 1990. Unidades limitadas por discontinuidades. Su aplicacion al Jurasico andino. Revista de la Asociacion Geologica Argentina, 45, 346-364. rieppel, o. 1989. A new pachypleurosaur (Reptilia: Sauropterygia) from the Middle Triassic of Monte San Giorgio, Switzerland. Philosophical Transactions of the Royal Society of London. Series B, 323, 1-73. seeley, H. G. 1874. Note on some generic modifications of the plesiosaurian pectoral arch. Quarterly Journal of the Geological Society, London, 30, 436-449. spalletti, L., gasparini, z. and Fernandez, m. 1994. Facies, ambientes y reptiles marinos de la transition entre las Formaciones Los Molles y Lajas (Jurasico medio), Cuenca Neuquina, Argentina. Acta Geologica Leopoldensia , 17, 329-344. storrs, G. 1991. Anatomy and relationships of Corosaurus alcovensis (Diapsida: Sauropterygia) and the Triassic Alcova Limestone of Wyoming. Bulletin of the Peabodv Museum of Natural History, 44, 1-151. stutchbury, s. 1846. Description of a new species of Plesiosaurus in the Museum of the Bristol Institution. Quarterly Journal of the Geological Society, London , 2, 41 1 — 41 7. sues, h.-d. 1989. Postcranial skeleton of Pistosaurus and interrelationships of the Sauropterygia (Diapsida). Zoological Journal of the Linnean Society, 90, 109-131. GASPARINI: BAJOCIAN PLIOSAUR 147 tarlo, l. b. 1960. A review of the Upper Jurassic pliosaurs. Bulletin of the British Museum ( Natural History ), Geology Series , 4, 147-189. taylor, M. a. \992a. Functional anatomy of the head of the large aquatic predator Rhomaleosaurus zetlandicus (Plesiosauria, Reptilia) from the Toarcian (Lower Jurassic) of Yorkshire, England. Philosophical Transactions of the Royal Society of London , Series B, 335, 247-280. - 1992 b. Taxonomy and taphonomy of Rhomaleosaurus zetlandicus (Plesiosauria, Reptilia) from the Toarcian (Lower Jurassic) of the Yorkshire coast. Proceedings of the Yorkshire Geological Society, 49, 49-55. and benton, M. J. 1986. Reptiles from the Upper Kimmeridge Clay (Kimmeridgian, Upper Jurassic) of the vicinity of Egmont Bight, Dorset. Proceedings of the Dorset Natural History and Archaeological Society, 107, 121-125. — and cruickshank, a. R. i. 1993. Cranial anatomy and functional morphology of Pliosaurus brachyspondylus (Reptilia: Plesiosauria) from the Upper Jurassic of Westbury, Wiltshire. Philosophical Transactions of the Royal Society of London, Series B, 341, 399^118. vignaud, p. 1995. Les Thalattosuchia, crocodiles marins du Mesozoi'que: systematique phylogenetique, paleoecologie, biochronologie et implications paleogeographiques. These Geosciences, Universite de Poitiers, 1-265. watson, D. M. s. 1910. Upper Liassic Reptilia. Part II, the Sauropterygia of the Whitby Museum. Memoirs and Proceedings of the Manchester Literary and Philosophical Society, 54, 1-13. welles, s. p. 1943. Elasmosaurid plesiosaurs with a description of new material from California and Colorado. Memoirs of the University of California, 13, 125-215. 1962. A new species of elasmosaur from the Aptian of Colombia, and a review of the Cretaceous plesiosaurs. University of California Publications in Geological Sciences , 44, 1-96. zhang yihong 1985. A new plesiosaur from Middle Jurassic of Sichuan Basin. Vertebrata PalAsiatica 23, 235-240. ZULMA GASPARINI Departamento Paleontologia Vertebrados Typescript received 5 September 1995 Museo de Ciencias Naturales de La Plata Revised typescript received 30 January 1996 1900 La Plata, Argentina APPENDIX: LIST OF ABBREVIATIONS a angular piv posterior interpterygoid vacuities adc anterodorsal crest pal palatine ar articular pmx premaxilla arpt anterior ramus of pterygoid ppr paroccipital process bo basioccipital prf prefrontal c coronoid ps parasphenoid d dentary Pt pterygoid ec ectopterygoid q quadrate en external naris qrpt quadrate ramus of pterygoid ep epipterygoid sa surangular fr frontal sp splenial ho hole sof suborbital fenestra in internal naris sq squamosal mx maxilla sqv squamosal ventral rami no notch stf subtemporal fenestra oc occipital condyle tf temporal fenestra P parietal V vomer pa prearticular THE DICYNODONT LYSTROSAURUS FROM THE UPPER PERMIAN OF ZAMBIA: EVOLUTIONARY AND STRATIGRAPHICAL IMPLICATIONS by g. m. king and i. jenkins Abstract. The skull of the dicynodont Lystrosaurus cf. curvatus is described from the Late Permian Madumabisa Mudstones of Zambia, in association with several Upper Permian genera. It demonstrates that the widespread Lystrosaurus, hitherto regarded as characteristic of the Lower Triassic, cannot be used in isolation as a biostratigraphical zone fossil. It appears that Lystrosaurus was a survivor of the Permo-Triassic extinction event, rather than a product of early Triassic diversification of other surviving forms. Its absence from the Upper Permian of South Africa suggests that it may have been an immigrant from further north. The Upper Permian fauna of the Madumabisa Mudstones is comparable to that of the Upper Guodikeng Formation of China. The fauna is younger than that of the Dicynodon Assemblage Zone of South Africa, but may be contemporaneous with that of the Cuttie’s Hillock Formation of Scotland. The anomodonts were a widespread, diverse and abundant group of mainly herbivorous therapsids (mammal-like reptiles) which lived in the Permian and Triassic (King 19906). A recent study of their generic diversity at their acme in the Late Permian Cistecephalus Assemblage Zone ( sensu Rubidge in press) of South Africa, suggested that approximately 15 genera were present (King 1993). This contrasts with the situation at the beginning of the Triassic (the Lystrosaurus Assemblage Zone) in South Africa, where only two genera Lystrosaurus and Myosaurus are known. Despite the paucity of genera in this assemblage, dicynodonts are nevertheless numerically abundant and hundreds of specimens have been collected from South Africa alone. The presence of more than one species of the genus Lystrosaurus is unusual for South African dicynodonts (King 1993), and could be linked to the dearth of other anomodonts of a similar size from the lowermost Triassic. The other genus of anomodont from the Lower Triassic, Myosaurus , was a very small animal, whereas Lystrosaurus was a medium-large anomodont. King (1991) and Cluver and King (1991) have shown that there is no firm evidence for the contention that Lystrosaurus was an aquatic or semi-aquatic animal, and so the reason for the success of Lystrosaurus is still to some extent a mystery. Specimens of Lystrosaurus have been reported from South Africa, India, China, Antarctica, Russia, possibly Australia and, more doubtfully, Laos. Their occurrence is taken to indicate an earliest Triassic age for the horizons in which they have been found, and the genus has therefore been considered to be a useful stratigraphical marker. In the South African Permo-Triassic Karoo Basin, the genus is a zone fossil for the Lystrosaurus Assemblage Zone (Rubidge in press) and is considered to be confined to that zone. In it, Lystrosaurus is found most commonly together with the anapsid Procolophon , the diapsid Proterosuchus , the therocephalians Moschorhinus , Scalopo- saurus and Regisaurus , and the cynodonts Thrinaxodon and Galesaurus. Kemp (1976) described a collection of therapsid fossils from the Madumabisa Mudstones of the Luangwa Valley in Zambia. The composition of this assemblage indicates a latest Permian ( Dicynodon Assemblage Zone) age for the localities. The following genera have been identified in the collection: Dicynodon (King 1981), Oudenodon (King 1979), Procynosuchus (Kemp 1979) and Diictodon (Gale 1988). Gorgonopsids and pareiasaur scutes are also present. A previously unidentified specimen in this collection is shown here to belong to the genus Lystrosaurus. This suggests either that Lystrosaurus occurs in the Upper Permian, or alternatively that several other genera of therapsids, hitherto regarded as Upper Permian, occur in the lowermost Triassic. It is [Palaeontology, Vol. 40, Part 1, 1997, pp. 149— 1 56| © The Palaeontological Association 150 PALAEONTOLOGY, VOLUME 40 more parsimonious to regard the Lvstrosaurus specimen as being from the Upper Permian. No other genera typical of the Lower Triassic have been found in the Zambian collection. Age of the Madumabisa Mudstones The presence of Lvstrosaurus in the Upper Permian of Zambia may indicate that the age of the Madumabisa Mudstones is very near the Permo-Triassic boundary, and possibly significantly later than the South African Dicynodon Assemblage Zone. It also indicates that a mixed Late Permian-Early Triassic fauna existed in Zambia, similar to the transitional fauna proposed by Cheng (1993) for the Upper Guodikeng Formation in China (see below). Some evidence for the greater age of the Zambian strata than those of South Africa is provided by the other dicynodonts of the Madumabisa Mudstones, in particular specimens of the genus Dicynodon. Several specimens in the Oxford University Museum TSK collection have a distinctive morphology which may indicate that they belong to a discrete species, probably D. trigonocephalus (King 1981). It is not possible to assert this with certainty, since the genus Dicynodon requires revision at the specific level. The distinctive features of these specimens are a medium-sized skull with breadth and length subequal giving a squarish dorsal profile; very abbreviated intertemporal region; short and deep basicranial region; and wide interorbital distance. The dicynodont from Knock of Alves, Elgin, in Scotland (Benton and Walker 1985, p. 209), although only adequately preserved in the snout region, bears a marked resemblance to the Zambian Dicynodon specimens in general shape and morphology (GMK, pers. obs.). Benton and Walker considered that the age of the Cuttie’s Hillock Formation at Elgin is also uppermost Permian. Walker (1973) tentatively suggested that the age might even be Lower Triassic. It would be interesting to determine whether Lystrosaurus Assemblage Zone strata are present in the Madumabisa Mudstones, and whether these might represent a continuous sequence through the Permo-Triassic boundary. Institutional abbreviations for specimens referred to in this work are as follows: BMNH, Palaeontological Collections, The Natural History Museum, London; OUM.TSK, T. S. Kemp Collection, Oxford University Museum. SYSTEMATIC PALAEONTOLOGY Subclass synapsida Osborn, 1903 Order therapsida Broom, 1905 Infraorder dicynodontia Owen, 1859 Superfamily pristerodontoidea Cluver and King, 1983 Family dicynodontidae Cluver and King, 1983 Subfamily kannemeyeriinae von Huene, 1948 Tribe lystrosaurini Broom, 1903 Genus lystrosaurus Cope, 1870 Diagnosis. Small to medium-sized dicynodont. Parietals widened; short snout, down-turned and deepened and formed by elongated maxilla and premaxilla; postcanine teeth absent; maxillary tusks present; orbits situated high on skull; nares immediately anterior to orbits; postfrontal present. Lystrosaurus curvatus (Owen, 1876) Broom, 1932 Holotype. BMNH R3792, a skull; Elandsburg, Cradock, Cape Province, South Africa; Lystrosaurus Assemblage Zone, Lower Triassic. KING AND JENKINS: PERMIAN DICYNODONT 151 text-fig. 1 . Lystrosaurus cf. curvatus\ OUM . TSK2 ; Late Permian Madumabisa Mudstones ; Luangwa Valley, Zambia. A, left lateral view, solid black shading indicates matrix; a large crack runs dorso-ventrally through the skull, b, palatal view, c, anterior view. Abbreviations: fr, frontal; nas, nasal; pmx, premaxilla. Scale bar represents 40 mm. Diagnosis. Snout not produced far ventrally; snout and tusk development weak; frontonasal and premaxillary ridges and frontal bosses absent; skull roof smooth; premaxillary plane curving over in a smooth arc to meet frontoparietal plane; frontal protuberances absent; suture between frontals and nasals lying in the frontoparietal plane; ventral ramus of squamosal extending posteriorly as well as laterally, concealing occipital condyles in lateral view. 152 PALAEONTOLOGY, VOLUME 40 Lystrosaurus cf. curvatus Text-figure 1a-c Material. OUM.TSK2, a skull. Locality and horizon. East side of hunter’s track from Luangwa River, along north side of Munyamadzi River, Luangwa Valley, Zambia; Madumabisa Mudstones, Upper Permian. Description. The specimen consists of a medium-sized skull (160 mm long) without the mandible (Text-fig. 1). The skull is more or less complete, but lacks the tip of the snout and the posterior zygomatic arches. The bone surface is slightly weathered. The following characters of Lystrosaurus can be seen: the basicranial axis is shortened posteriorly; the parietals in the intertemporal bar are wide and not covered completely by the postorbitals; the snout is bent downwards making an angle with the skull roof, and is deepened (King 19906). The suture between the premaxilla and maxilla is smooth, and the premaxilla extends posterodorsally as far as the prefrontals (pmx. Text-fig. Ic); the external naris is pear-shaped and bears a rugose ridge at its posteroventral edge (King 1991 ; King and Cluver 1991). Caniniform tusks are present. Remarks. Of the species described by Cluver (1971) and CosgrifF et al. (1982), OUM .TSK2 appears most similar to Lystrosaurus curvatus in having a smoothly curved skull roof in profile, and lacking a fronto-nasal ridge, ornament on the frontals, prefrontal bosses, and laterally flared squamosals. Cluver considered L. curvatus to be the most primitive of the Lystrosaurus species and this would be consistent with its presence in the Upper Permian. It might be questioned whether this relatively small specimen might not simply be a juvenile of the Dicynodon species present in the OUM.TSK collection. As noted below, there are other specimens present (probably belonging to Dicynodon trigonocephalus) which have a shortened and deepened basicranial axis (as in Lystrosaurus), but very narrow ridge-like intertemporal regions which would exclude them from the genus Lystrosaurus. It is possible that the wider intertemporal region of the small so-called Lystrosaurus specimen is a juvenile feature, becoming narrower during ontogeny. Against this proposition is the fact that the premaxillae of the two forms are quite distinctive. In OUM.TSK2, the premaxilla extends proportionately further dorso-posteriorly, almost separating the nasals, while its suture with the nasals is smooth-edge, not interdigitating as in the Dicynodon specimens (Text-fig. 1 c). King ( 1991 ) found this to be a consistent and functionally important feature in Lystrosaurus. DISCUSSION Transitional Lystrosaurus from other regions Although it is currently agreed that the presence of Lystrosaurus indicates lowermost Triassic age (e.g. Cosgriff et al. 1982; Olson 1989), the fossil has been reported previously to be found in association with typical members of Upper Permian faunas. Hotton (1967) described a section of the Lystrosaurus Assemblage Zone at Lootsberg Pass, Orange Free State, South Africa in which he noted that specimens of Lystrosaurus overlapped in the section for about 60 m (200 feet) with typical members of the underlying Dicynodon Assemblage Zone fauna such as Daptocephalus (= Dicynodon) and Moschorhinus. Because the latter were found in differently coloured shales from those containing Lystrosaurus, Hotton postulated that two different contemporaneous facies were present, and that Moschorhinus and Daptocephalus were conservative Dicynodon Assemblage Zone forms which had survived into the Lystrosaurus Assemblage Zone, perhaps in different niches from those occupied by Lystrosaurus. Kitching (1977) mentioned that in areas with some geographical relief, Daptocephalus has often been recorded from the Lystrosaurus Assemblage Zone. This has been in situations where the Lystrosaurus Assemblage Zone fauna accompanying it could not have been washed or rolled down from higher strata. Similarly, Lystrosaurus has also been recorded from the Dicynodon Assemblage Zone. Kitching, however, considered such occurrences to represent the circumstance of Lystrosaurus KING AND JENKINS: PERMIAN DICYNODONT 153 Assemblage Zone sediments having been laid down in previously existing erosional channels of the underlying Dicynodon Assemblage Zone rocks, notably in the Lootsberg area. Kitching did not consider this association to represent true contemporaneity of the faunas. However, contemporaneous faunas do appear to be present in the Upper Permian of China. Olson (1989) noted that mixed Dicynodon-Lystrosaurus faunas had been reported from the Guodikeng Formation of Xinjiang, China; and these reports have subsequently been discussed in more detail by Cheng (1993). The earliest Triassic Jiucaiyuan Formation contains specimens of a large species of Lystrosaurus, whereas the lower and middle zones of the Late Permian Guodikeng Formation contain the typical Permian anomodont Striodon. The Upper Guodikeng Formation, however, contains a small species of Lystrosaurus and the typical Permian anomodont Jimusuaria. Cheng considered the Upper Guodikeng Formation to contain a transitional, continuous Late Permian-Early Triassic fauna. The pollen assemblage from the Upper Guodikeng is also of a transitional nature. This situation seems to be very similar to that of the Late Permian Madumabisa Mudstones in Zambia. Cheng (1993) stressed the significance of the discovery of the transitional zone in China with respect to Permo-Triassic stratigraphy and evolution. The Zambian strata are potentially even more important because of the wealth of extremely well-preserved vertebrate fossils they contain. However, an alternative explanation may be that both the Madumabisa Mudstones and the Upper Guodikeng strata are Upper Permian rather than transitional, as evinced by the presence of Striodon and Jimusuaria in the Chinese localities, and the similarity of the Zambian Lystrosaurus specimen to Dicynodon trigonocephalus. Stratigraphical use of Lystrosaurus The occurrence of Lystrosaurus in Late Permian rocks indicates that isolated specimens of the genus should no longer be used for biostratigraphical purposes. Unless other Triassic genera were to be found with the Late Permian ones, it remains reasonable to use an assemblage of genera, of which Lystrosaurus is part, to correlate lowermost Triassic rocks, but use of Lystrosaurus alone could be misleading. This is obviously unfortunate, since Lystrosaurus is the most common genus in many assemblages and so most likely to be encountered in the course of stratigraphical work. Survivorship of Lystrosaurus and the end-Permian extinction event No other Permian anomodont is known to cross the Permo-Triassic boundary, so why did Lystrosaurus survive the end-Permian event when so many other genera did not? One possibility is that Lystrosaurus was adapted to feeding on some component of the transitional flora which succeeded the Glossopteris flora of the southern hemisphere towards the end of the Permian (Tucker and Benton 1982). If the food-plant utilized by Lystrosaurus became common in the Dicroidium flora of the Late Permian-Early Triassic, this could explain the rise in abundance of Lystrosaurus. In favour of this, at least as a partial explanation, is the change in organization of the feeding system seen in Lystrosaurus. Whereas in many Permian dicynodonts (e.g. Diictodon , Oudenodon , Robertia) the backward-pulling component of the external adductor muscles was very substantial and produced longitudinal movement of the lower jaw, in Lystrosaurus this component was reduced (King 19906; Cox 1991; King and Cluver 1991). The external adductor muscles had a greater vertical component in the latter and so produced a more strictly orthal jaw movement. King and Cluver (1991) have argued that Lystrosaurus was adapted to feeding on resistant vegetation, and had specializations of the skull to deal with this. It is impossible to ascertain which component(s) of the transitional or Dicroidium floras might have been the relevant food source. Further support for this argument is provided by the observation that the Mesophytic flora containing Dicroidium replaced the G/oYso/?tera-dominated Palaeophytic flora in a north-south sequence through South Africa during the Late Permian to Early Triassic (Andrews 1961). This appears to have coincided with the extinction of numerous anomodont genera and also the 154 PALAEONTOLOGY, VOLUME 40 migration of Lystrosaurus from northern Zambia to South Africa. Although the origin of the Mesophytic flora has been traced to the Late Carboniferous (DiMichele and Aronson 1992; Erwin 1993) with the transition beginning at low latitudes and spreading towards the poles, the final transition did not occur until the Early Triassic. This suggests that Lystrosaurus utilized some components of the new Mesophytic flora, either as a specialist or a generalist, in contrast to other anomodont genera which presumably could not feed on the ‘new’ plant types. The occurrence of Lystrosaurus in the Lower Triassic and not in the Upper Permian has been one of the most clearly visible pieces of evidence for a terrestrial mass extinction event at the Permian-Triassic boundary. However, if Lystrosaurus was present in the Late Permian, both the taxonomic extent and the suddenness of the extinction event are brought into question. The disappearance of so many Permian genera from the fossil record prior to the end of the period is usually interpreted as a consequence of some rapid and drastic environmental change. Smith’s (Smith 1990; Smith et al. 1993) reviews of Permo-Triassic palaeoenvironments illustrate the increasing aridity of the Southern African area occurring in a north to south direction through the later half of the Permian and into the Triassic. The changes occurred as a result of the northward movement of Africa. The Teekloof Formation of South Africa, deposited contemporaneously with the Madumabisa Mudstones, shows a palaeoenvironment that is already semi-arid with highly seasonal rainfall (Smith et al. 1993). Hotton (1986) noted that striae on the medial wear facet of a tusk in a specimen of Lystrosaurus from the Karoo are partly obliterated by polishing. He suggested that this might have been caused by a change in the nature of its food in association with altered feeding circumstances. He postulated the seasonal alternation of harsh and succulent plant material, or alternatively subsurface and above-ground plant elements as an influencing factor. These findings might also suggest that increasing aridity and seasonality of rainfall, accompanied by a major floristic change, had a significant deleterious effect on Late Permian anomodonts, except for Lystrosaurus which appears to have been capable of existing in such environments. Erwin's (1994) suggestion of global warming from oceanic anoxia via increased atmospheric carbon dioxide also lends some support to a model of increasing environmental harshness in southern Gondwana. However, the implication from these observations is that the environmental changes occurred gradually. Vacant niches in the changing environment after the event are assumed to have been filled by new radiations centred on surviving taxa. Lystrosaurus had previously been interpreted as having evolved as part of a post-extinction radiation, thus constituting part of the evidence for turnover at the Permian-Triassic boundary. However, the presence of Lystrosaurus in the Upper Permian of Zambia indicates that it was not part of any new adaptive radiation. This may be true of other members of the Lystrosaurus Assemblage Zone fauna, but none have yet been detected in the Permian. If others did exist in the Permian, the evidence for a rapid and taxonomically widespread terrestrial Permian extinction would be reduced. King (1990n) has already questioned the suddenness of the event in the Karoo Basin, as there is evidence that several groups of tetrapods were declining in diversity before the end of the Permian. The existence of a supposed post-extinction genus before the event throws more doubt on its nature. Geographical origin of Lystrosaurus In 1977, Kitching (p. 23) commented that ‘The sudden appearance of this new form of anomodont, Lystrosaurus , and its abundance throughout the zone is more indicative of an immigrant form than of one evolved from a branch of the anomodonts from the lower zones. Had the genus Lystrosaurus been evolving from a dicynodont from the lower zones then it is considered that corroborative evidence should have been found among some of the large variety of Cistecephalus and Daptoceplmlus Zone dicynodonts’. He therefore considered that the origin of Lystrosaurus in the Karoo Basin must have been by immigration since there is no evidence from the underlying strata that this dicynodont evolved in situ. This is not the case with the Permian strata in Zambia. Not only is Lystrosaurus now known from the Upper Permian, but there are also other KING AND JENKINS: PERMIAN DICYNODONT 155 dicynodont genera which resemble it, in particular Dicynodon trigonocephalus. In describing this Zambian form. King ( 1981 ) mentioned that it had certain features reminiscent of Lystrosaurus, such as the medium-sized skull with subequal length and breadth giving a squarish dorsal profile; very short intertemporal region; short and deep basicranial region; and wide interorbital distance. Several Zambian skulls with this morphology are present in the OUM.TSK collection. These characters suggest that this species of Dicynodon is the sister-taxon of Lystrosaurus and could imply that Lystrosaurus did evolve in situ in the Zambian Basin in the Late Permian and migrated into the Karoo Basin subsequently. This underlines the effect that migrations may have upon the observed pattern of faunal change in a discrete area. An alternative possibility pointed out by A. R. Milner (pers. comm.) is that since the Madumabisa Mudstones are younger than the Dicynodon Assemblage Zone and older than the Lystrosaurus Assemblage Zone, they could belong with the interregnum between these two zones. Thus, Lystrosaurus may have evolved within this interregnum in South Africa, its sudden appearance resulting from the magnitude of the temporal interval in the fossil record, this disunion being partly filled by the Madumabisa Mudstones in Zambia. Acknowledgements. Thanks go to Tom Kemp for permission to work on the specimen, Michael Cherry, Michael Cluver and Andrew Milner for valuable comments, Cedric Hunter for artwork, and St John’s College, Oxford and the Sedgwick Museum, Cambridge for hospitality. REFERENCES Andrews, h. n. 1961. Studies in paiaeobotany . Wiley, New York, 487 pp. benton, M. J. and walker, a. D. 1985. Palaeoecology, taphonomy, and dating of Permo-Triassic reptiles from Elgin, north-east Scotland. Palaeontology , 28, 207-234. broom, r. 1903. On the classification of the theriodonts and their allies. Report of the South African Association for the Advancement of Science , I, 286—295. 1905. On the use of the term Anomodontia. Record of the Albany Museum , 1, 266-269. — 1932. The mammal-like reptiles of South Africa and the origin of mammals. H. F. and G. Witherby, London, 376 pp. cheng zheng-wu 1993. On the discovery and significance of the nonmarine Permo-Triassic transition zone at Dalongkou in Jimusar, Xinjiang, China. 65-67. In lucas, s. g. and morales, m. (eds). The nonmarine Triassic. Bulletin of the New Mexico Museum of Natural History and Sciences , 3, 1-478. cluver, m. A. 1971. The cranial morphology of the dicynodont genus Lystrosaurus. Annals of the South African Museum , 56, 155-274. and king, G. M. 1983. A reassessment of the relationships of Permian Dicynodontia (Reptilia, Therapsida) and a new classification of dicynodont. Annals of the South African Museum , 91, 195-273. cope, e. d. 1870. On the skull of dicynodont Reptilia. Lystrosaurus frontosus from Cape Colony. Proceedings of the American Philosophical Society , 11, 419. cosgriff, j. w., hammer, w. r. and ryan, w. J. 1982. The Pangaean reptile, Lystrosaurus maccaigi in the Lower Triassic of Antarctica. Journal of Paleontology. 56, 371-385. cox, c. b. 1991 The Pangaean dicynodont Rechnisaurus and the comparative biostratigraphy of Triassic dicynodont faunas. Palaeontology , 34, 767-784. dimichele, w. a. and aronson, r. b. 1992. The Pennsylvanian-Permian vegetational transition: a terrestrial analogue to onshore-offshore hypothesis. Evolution. 46, 807-824. erwin, d. H. 1993. The great Paleozoic crisis: life and death in the Permian. Columbia University Press, 327 pp. 1994. The Permo-Triassic extinction. Nature , 357, 231-235. gale, t. m. 1988. Comments on a 'nest' of juvenile dicynodont reptiles. Modern Geology , 13, 119-124. hotton, n. in 1967. Stratigraphy and sedimentation in the Beaufort Series (Permian-Tnassic), South Africa. 380-428. In teichert, c. and yochelson, e. l. (eds). Essays in paleontology and stratigraphy. Raymond C. Moore commemorative volume. University of Kansas Special Publication, 2. University of Kansas Press, Lawrence, 626 pp. - 1986. Dicynodonts and their role as primary consumers. 71-82. In hotton, n. hi, maclean, p. d., roth, J. J. and roth, e. c. The ecology and biology of mammal-like reptiles. Smithsonian, Washington and London, 326 pp. 156 PALAEONTOLOGY, VOLUME 40 huene, f. von 1948. Short review of the lower tetrapods. 65-106. In Royal Society of South Africa Robert Broom Commemorative Volume. Royal Society of South Africa, 257 pp. kemp, t. s. 1976. Vertebrate localities in the Karroo system of the Luangwa Valley, Zambia. Nature , London , 254, 4 1 5 — 4 1 6 . — 1978. The primitive cynodont Procynosuchus : functional anatomy of the skull and relationships. Philosophical Transactions of the Royal Society of London , Series B, 285, 73-122. king, G. M. 1979. Permian dicynodonts from Zambia. Unpublished D.Phil. thesis, University of Oxford. 1981. The functional anatomy of a Permian dicynodont. Philosophical Transactions of the Royal Society of London, Series B, 291, 243-322. I990u. The dicynodonts , a study in palaeobiology. Chapman and Hall, London, 233 pp. 19906. Dicynodonts and the end Permian event. Palaeontologia africana , 27, 31-39. 1991. The aquatic Lystrosaurus: a palaeontological myth. Historical Biology, 4, 285-321. 1993. Species longevity and generic diversity in dicynodont mammal-like reptiles. Palaeogeography , Palaeoclimatology, Palaeoecology, 102, 321-332. — and cluver, m. a. 1991 . The aquatic Lystrosaurus : an alternative lifestyle. Historical Biology, 4, 323-341 . kitching, j. w. 1977. Distribution of the Karroo vertebrate fauna. Memoirs of the Bernard Price Institute for Palaeontological Research, 1, 1-131. olson, e. c. 1989. Problems of Permo-Triassic terrestrial vertebrate extinctions. Historical Biology, 2, 17-35. osborn, h. f. 1903. On the primary division of the Reptilia, into two sub-classes, Synapsida and Diapsida. Science , 17, 275-276. owen, R. 1859. On the order of fossil and recent Reptilia, and their distribution in time. Report of the British Association for the Advancement of Science, 1859, 153-166. — 1876. Descriptive and illustrated catalogue of the fossil Reptilia of South Africa in the collection of the British Museum. British Museum (Natural History), London, 88 pp. RUBIDGE, B. s. (ed.) in press. Biostratigraphy of the Beaufort Group (Karoo Supergroup), South Africa. Biostratigraphic Series No. 1. South African Commission for Stratigraphy, Council for Geoscience, Pretoria, 46 pp. smith, r. m. h. 1990. A review of stratigraphic and sedimentary environments of the Karoo Basin of South Africa. Journal of African Earth Science, 10, 1 17-137. — eriksson, p. G. and botha, w. j. 1993. A review of the stratigraphic and sedimentary environments of the Karoo-aged basins of South Africa. Journal of African Earth Science, 16, 143-169. tucker, m. E. and benton, m. j. 1982. Triassic environments, climates and reptile evolution. Palaeogeography, Palaeoclimatology, Palaeoecology, 40, 361-379. walker, a. D. 1983. The age of the Cuttie’s Hillock Sandstone (Permo-Triassic) of the Elgin area. Scottish Journal of Geology, 9, 177-183. YANG JI-DUAN, QU LI-FAN, ZHOU HUI-QI, CHENG ZHENG- WU, ZHOU TONG-SHUN, HOU JING-PEN, WU XAO-ZU, ZHANG zhi-ming and wang zhi 1984. Permian and Triassic strata and fossil assemblages in the Dalongkou area of Jimusar, Xinjiang. Memoirs of the Institute of Geology of the Chinese Academy of Geological Science, 2(3), 1-262. G. M. KING Faculty of Classics University of Cambridge Sidgwick Avenue Cambridge CB3 9DA, UK formerly Division of Earth Sciences South African Museum PO Box 61, Cape Town 8000 South Africa Typescript received 3 May 1994 Revised typescript received 12 March 1996 I. JENKINS Department of Earth Sciences University of Cambridge Downing Street Cambridge CB2 3EQ, UK FIRST RECORD OF FOOTPRINTS OF TERRESTRIAL VERTEBRATES FROM THE UPPER PERMIAN OF THE CIS-URALS, RUSSIA by VALENTIN P. TVERDOKHLEBOV, GALINA I. T VERDOKH LEBO V A, MICHAEL J. BENTON and GLENN W. STORRS Abstract. The first tetrapod footprints from the Upper Permian of Russia are identified as Anthichnium ichnosp., on the basis of a short track from the Severodvinskian Gorizont (upper Tatarian, uppermost Permian) of Kulchomovo, 75 km east-north-east of Orenburg, southern Pre-Urals basin. The footprints are preserved in a ripple-marked sandstone unit, and appear to show an animal swimming, then crawling through wet marginal sediment, and finally moving on to firmer substrate. The sedimentary setting is an enclosed semi- arid foreland basin, close to high mountains. An ephemeral stream-playa lake complex developed, with long emergent periods and short episodes of flooding and stream development. The footprints were probably produced by a small temnospondyl amphibian with four fingers and five toes. More than 1000 occurrences of terrestrial vertebrates are known from the Upper Permian and Triassic continental red beds of the easternmost European part of Russia (Efremov and Vjushkov 1955; Olson 1957; Tverdokhlebova 1976; Ochev et at. 1979). Tetrapod remains have been found in virtually all areas where such strata occur, from the Barents Sea in the far north, well within the Arctic Circle, to the Pre-Caspian region in the south (Text-fig. 1). However, vertebrate tracks have not hitherto been found in this vast territory, and we make the first report here of such an occurrence. The footprints, dated as Tatarian (Upper Permian), are small, and the track is preserved on a series of seven adjoining blocks that are easily transported and pieced together. These are housed in the collection of the Scientific Research Institute of Geology, Saratov State University, SGU No. 104 B/2060-2065. Four additional and adjacent blocks without tracks are also preserved. GEOLOGICAL SETTING Locality and age The footprints were found in 1988 by one of us (VPT) in the southern Cis-Uralian (Pre-Ural) Trough close to the right bank of the Sakmara river, 75 km east-north-east of the city of Orenburg (Text- fig. 1). The locality lies just to the north of a graded dirt road (‘graida’) on the west bank of a small unnamed stream (Text-fig. 2), about 12 km north-west of Saraktash. The ‘graida’ runs east-south- east from the village of Gavrilovka and crosses the stream in question after about 10 5 km, and about 1-5 km before the road passes the tiny settlement of Kulchumovo. The map reference for the locality, using the Soviet ‘Sistema Koordinat 1942’ is 10449257505. The longitude of the site is 56° 16' E, and the latitude is 51° 53' N. The sedimentary rocks exposed in the stream section belong to the Severodvinskian (North Dvina) Gorizont (‘horizon’), of late Tatarian (latest Permian) age. This unit is a widespread, and widely recognized, biostratigraphical division in both the Ural and Povolzhye (near-Volga) regions (Anfimov et al. 1993). A Russian gorizont is not equivalent to a lithostratigraphical horizon of western nomenclature, but rather is defined solely on the basis of its contained fauna and flora. The Severodvinskian is characterized especially by the ostracod Suchonellina futschiki. It has yielded also (Palaeontology, Vol. 40, Part 1, 1997, pp. 157-166| © The Palaeontological Association 158 PALAEONTOLOGY, VOLUME 40 PERMO-TRIASSIC TETRAPOD FAUNAS Main localities O Bukobay 0 Donguz + Yarenga ^ Upper Vetluga X Lower Vetluga ^ Vyazniki ^ North Dvina O Kotelnich SJ Isheyevo Mezen-Belebey | Ocher text-fig. 1. Map of major Permo-Triassic tetrapod localities on the Russian Platform and in the Cis-Uralian Trough. The specimens described herein are from near Orenburg, in the south-east. the bivalves Palaeomutela orthodonta, P. ovalis, and Sacmariella novovulvhumica , the gastropods Vetlugaia aristovensis and V. suchonensis , the conchostracans Pseudestheria plotnikovi and Sphaerestheria belorussica , and the ostracods Darwimda inornata , D. inornata var. macra, and D. paral/ela (Anfimov et al. 1993). Associated vertebrate fossils include skeletons of temnospondyls (Dvinosawus egregius, D. primus ), anthracosaurs (Bystrowiana permira , Chroniosaurus dongusensis, Jugosuchus boreus , J. hartmanni , Rapbanodon tverdochlebovae ), pareiasaurs ( Deltavjatka vjatkensis , Scutosaurus permianus, S. itilensis, S. rossicus), dicynodonts ( Dicynodon trautscholdi), gorgo- nopsians (Probumetia vjatkensis , Sauroctonus progressus ) and therocephalians ( Moschowhaitsia vjuschkovi, Niuksenitia sukhonensis) (Tverdokhlebova 1976). TVERDOKHLEBOV ET AL.\ PERMIAN FOOTPRINTS 159 56°00' 10436 40 44 48 52 56 60 64 The Severodvinskian Gorizont is the lower of two biostratigraphical divisions forming the Upper Tatarian Substage, the higher one being the Vyatskian Gorizont. Below the Severodvinskian, the Urzhumskian Gorizont corresponds to the entirety of the Lower Tatarian Substage. In general, the biostratigraphical divisions of the Upper Permian in Russia, and especially the Tatarian, are based upon fossil tetrapod data (Efremov and Vjushkov 1955; Tverdokhlebova 1976), although recent studies of ostracods have produced good section correlations (Molostovskaya 1982, 1993). The utility of bivalves and conchostracans is as yet limited. It may be noted here that Kozur (1993) has placed the Cis-Uralian Tatarian rather earlier than latest Permian. Cis-Uralian rocks belonging to the Severodvinskian are seen particularly in the north-west part of the Perm region, some 600 km north of Orenburg. Around Perm, they consist mainly of limestones, some of which are dolomitized, and marls. Some of the limestones show evidence of algal/stromatolitic origins. East and south of Orenburg, Severodvinskian rocks consist of cyclic sequences of alternating mudstones, siltstones, and flaggy sandstones, with bedded clays, limestones, and conglomerates (Garyainov et al. 1967, p. 4). The lithostratigraphical unit of the Kulchumovo locality is unnamed locally, but may be equivalent to the Severodvinskian Malokinelsk Svita of the Russian Platform. A Russian svita is the nearest lithostratigraphical equivalent to a western formation, although the definitions of such svitas are less formal. 160 PALAEONTOLOGY, VOLUME 40 Sedimentology The footprints were found on the surface of a light-grey fine-grained calcareous sandstone, which is laterally discontinuous, and up to 0-05 m thick. The sandstone unit is one of many in a sequence (Text-fig. 3) of mudstones and sandstones, in which mudstones make up about 90 per cent, of the continued (right) continued (right) covered continued (nght) covered I H covered red mudstone grey laminated mudstone sandstone (AA/4 ripple marks ** plant debris TV" mudc racks K y burrows 1 vertebrate footprints 1 m text-fig. 3. Logged stratigraphical section along unnamed stream bed near Kulchumovo containing tetrapod trackway. The locations of parts of the logged section are shown in the sketch-map at top right. thickness of the succession. Some of the thin sandstone beds extend laterally for 10 m or more, but there are some obvious channels, such as at point E in the logged section (Text-fig. 3). The succession consists of 70 m or more of mudstones, siltstones, and sandstones, showing ripple-cross- lammated sets 0-2 m thick. The sandstones are grey-buff; the mudstones may be grey or red. The mudstones are all laminated to some extent, but the lamination is clearest in the grey mudstones. TVERDOKHLEBOV ET AL.\ PERMIAN FOOTPRINTS 161 Fossils in these units consist of carbonized vegetation and feeding burrows of invertebrates, as well as conchostracans, ostracods, bivalves and macerated fish remains. The sandstone slabs which bear the tracks vary in thickness, being thinnest towards the bottom. They are grey, of very fine sand or silt grade and are only slightly laminated, almost massive. A carbonate cement binds the grains. They are considered to have been deposited subaqueously, as indicated by the nature of the tracks and associated sedimentary structures (see below). The slabs bear current ripple marks on their top surfaces. These ripples are lobate and semi-crescentic with sharp asymmetrical crests, up to c. 8 mm high. Flow direction was from top to bottom, as illustrated. Ripple cross-laminations occur also, on a scale of 50-80 mm long and 12-20 mm high. Obvious rain-marks are seen all over the upper surface of the slabs, clearly developed on ripple crests, but much less distinct or absent in the ripple troughs. All have indistinct borders. One slab shows four successive rain-print levels in cross section. The surface troughs between ripples show evidence of having been filled with rain water, which later drained away and/or evaporated, leaving successive, finely defined, concentric tide marks. The slabs lay at the top of a unit of red mudstones. On the under surface of the footprint-bearing sandstones there are impressions of crescentic ripples, desiccation cracks and rain-drop prints, cast from the underlying unit. The desiccation cracks are 1 -5-2-0 mm deep. SYSTEMATIC PALAEONTOLOGY Genus anthichnium Nopcsa, 1923 Anthichnium ichnosp. indet Text-figure 4 Description The tracks are those of a small (approximately 100 mm long), sprawling, quadruped. The prints are generally indistinct, but the overall pattern of sinuous movement is clear. The track proceeds from the bottom to the top of the specimen as figured (Text-fig. 4). The first marks, on slabs 1-4 (Text-fig. 4), are in the form of furrows made by claws and traces of pads, seemingly produced by a floating animal whose feet barely touched bottom. However, these marks are very distinct since they were imprinted on a fine carbonate mud film overlying the typical sand. Farther up the track (slabs 3, 5), where the animal stepped with its full weight on the bottom, traces left by its longer hind limbs are evident (e.g. Text-fig. 4, print a). These have rather smooth and rounded edges, presumably since the sediment near the water’s edge was oversaturated or even covered with water. In this region, however, there are distinct rain-drop imprints on the landward side of the ripple crests (slabs 4-5). The 'land’ was wet fine-grained sand arranged in cuspate ripple marks, with small pools of water lying in the depressions between ripples. After passing out of the water-saturated littoral zone on to firmer substrate (slabs 5-7), the footprints become better defined, and the toe marks are clearly seen, some terminating in pointed claw marks. Near the top of slab 5, the animal climbed the stoss side of a ripple, and three clear footprints are preserved on the crest (Text-fig. 4, prints c-d). It is interesting that the leftmost two of these prints (c) appear to represent the same foot, a second successful attempt to climb the ripple following an initial slip of the left foot (print c) from the muddy ripple crest. Print d lies on the downstream edge of the crest and the heel impression is not preserved. Above that site, on slabs 6 and 7, the animal followed a flatter route, and produced more regular tracks (prints e-i. Text-fig. 4). Print 6e (Text-fig. 4) exhibits short slippage marks behind the heel. Prints f-h have more splayed toes in somewhat deeper mud; a ridge of mud has been pressed up by the step forming print 6h (Text- fig. 4). A similar extruded mud crest lies to the right of the middle (right hind) print of slab 5 (print b. Text- fig. 4). The track swings to the right near the top of slab 7 along the line of the ripple crests. Most of the footprints are incomplete, having washed-away or water-worn edges. The best preserved hindprints, on block 6 (Text-fig. 4), measure approximately 8 mm wide x 10 mm long. Their five toes are much longer and more slender than the digits of the forelimbs; digit 4 is apparently the longest. The best foreprint, on slab 7 (print j. Text-fig. 4), is c. 6 x 6 mm and shows only four short and stubby digits, number 3 being the longest. None of the other less well preserved foreprints shows more than four digits. The smoothed, slipping, footprints in the ‘shoreline’ region (slabs 4 and 5) are 5-7 mm long, and 1 • 5—2 0 mm deep. The spacing between opposite limbs was 25-30 mm, based on measurements of maximum track width. The traces left by fore and hindlimbs generally occur close together, or even overlap, with the hind occasionally stepping in and 162 PALAEONTOLOGY, VOLUME 40 text-fig. 4. Schematic interpretation of assembled tetrapod trackway slabs, Anthichnium ichnosp. (SGU No. 104 B/2060-2065), from the late Tatarian, Severodvinskian Gorizont, near Kulchumovo, Orenburg Oblast, Russia, and detail of four of the prints. Waterline indicated by arrows. Scale bars represent 100 mm (main slab), and 10 mm (individual prints b, e, h, j). TVERDOKHLEBOV ET AL.: PERMIAN FOOTPRINTS 163 obliterating the foreprint, and is typical for the trackway of a quadrupedal sprawler. Stride length varies enormously within this short segment of trackway, as the animal's progress was variously impeded by slick substrate conditions and topographical obstacles. For instance, the strong ripples of block 7 interfered with the creature’s movement and significantly shortened its stride. The stride length appears most uniform and normal on the relatively high, hard ground of slab 6 (Text-fig. 4). Here it may also be measured most readily, where three clear impressions of the left hindlimb (e, g, i) are spaced, heel to heel, 47 mm (e to g) and 41 mm (g to i) apart. Two imprints of the right hindlimb (f, h), also on slab 6, indicate a heel to heel stride length of 46 mm. The foreprints lie on average 4-6 mm in front of the hindprints on block 6. Identification of the trackmaker Comparison with previously described footprints suggests one of three tetrapod groups as trackmaker: amphibians, procolophonids, and diapsids, all of which are well known from the Upper Permian. Of these, the procolophonids and diapsids generally had five fingers and five toes, and footprints ascribed to those two groups (such as Palmichnus or Phalangichnus, and Anhomoiichnium respectively) are too large for comparison (Haubold 1971, pp. 31, 43-44). Small amphibian footprints, such as Anthichnium and Batrachichnus , on the other hand, are comparable. Individual prints of these taxa are generally 10 mm, or less, in length, the foot has longer digits than the hand, there are four fingers and five toes, and the foot is plantigrade (Haubold 1971, pp. 13-14). Both have generally been reported only from the Upper Carboniferous and Lower Permian (Haubold 1971; Fichter 1983), but excellent tracks identified as Anthichnium have been reported from the Upper Permian of Provence, France (Demathieu et al. 1992; Gand et al. 1995). These show all the features noted above, as well as the relative lengths of the toes, which match those seen in the Russian specimen (b, e. Text-fig. 4). Gand et al. (1995) classify Anthichnium in the temnospondyl family Branchiosauridae, but that assignment is not certain. Among amphibians from the Severodvinskian Gorizont of the Cis-Urals area, the anthracosaurs Chroniosaurus and Raphanodon were small enough to be the trackmaker. The digit counts and phalangeal formulae of these forms are unknown, but it is unlikely that they would have had four manual digits. Temnospondyls, on the other hand, had four digits. The temnospondyls from the Severodvinskian Gorizont of the Cis-Uralian area, species of Dvinosaurus, were too large in comparison with the tracks. However, juvenile Dvinosaurus , or an as yet unknown adult temnospondyl species, was probably responsible. DISCUSSION Sedimentary environment The sedimentary succession exposed at the trackway locality near Kulchumovo appears to represent a marginal lake environment, probably of ephemeral nature, in the broad shallow basins to the west of the ancestral Urals. The mix of grey and red mudstones with minor fine-grained calcareous sandstones, many of them laterally continuous, the associated ripple marks, rain-drop impressions, desiccation cracks, disseminated plant debris, and vertebrate tracks are all typical features of such environments (Picard and High 1972). The trackway slabs represent a narrow zone close to the edge of a pool or lake, into which rainwaters flowed periodically, creating the lobate, asymmetrical ripples. The setting is similar to that of the Late Permian Lower Beaufort Group of the Karoo Basin, South Africa, the environment of deposition of which is interpreted (Stear 1983) as an ancient ephemeral stream-playa lake complex that formed in a semi-arid inland basin. Sheet-like and lenticular sandstone bodies occur as superimposed systems of fluvial channels and overbank splays, with abundant evidence for falling water levels and emergence. Stear (1983) interpreted the Lower Beaufort Group sediments as largely high-energy clastic materials washed into a semi-arid basin from seasonal rainfall or glacial meltwaters in surrounding upland areas. These periodic short-term inundations incised deep channels, and waters ran to the middle of the basin where there was a 164 PALAEONTOLOGY, VOLUME 40 shallow lake or playa. The water seeped into the ground or evaporated during the rest of the year, when rainfall or glacial runoff was minimal, and emergent features, such as desiccation cracks, evaporites, and calcretes, developed during these times. The Pre-Ural basin shares all of these sedimentological features, and its tectonic setting (foreland basin) is also the same as the South African example. Fine- and coarse-grained sediment transported from arid plains into lakes that lack any drainage route present highly favourable conditions for preservation of surface sedimentary structures and footprints. The Kulchumovo sediments, however, indicate a hot, semi-arid climate with intense seasonality. There is no evidence locally for dramatic rainfall which would have produced sudden flooding events, forming deep wadi-like channels. If such wadis existed, they were presumably situated more proximally, on relatively high mountain slopes. Rain showers were apparently moderate locally (although potentially torrential proximally), raising water levels periodically, and bringing silt and sand pulses into lakes and pools that normally experienced slow deposition from suspension of clays and muds. Between rain showers, there were drying phases, during which the exposed mud surfaces became cohesive. Further rainfall brought in more sediment rapidly enough to preserve surface impressions and desiccation cracks, while continuing the cycle of lake level oscillation. Tverdokhlebov (1988, 1989) has suggested that regional conditions allowed for deposition of great thicknesses of mud in a relatively short period of time, perhaps seasonally. The change at the footprint level from red to grey rocks corresponds to a locally more stable aquatic environment and reducing conditions. Fossils of aquatic organisms are more common above the horizon that yielded the tracks. The Late Permian sequences of the Cis-Urals are interpreted (Newell et al. in press) as deposits of a transverse drainage system characterized by channels which decreased in size downcurrent and terminated in a muddy flood basin. The range of facies indicates a terminal fan setting. The Kulchomovo locality belongs to Newell et al.' s (in press) distal facies association: deposits of a muddy basin which generally received only fine-grained sediments after large floods. Occasional sheet sandstones and heterolithic channel fills are incised into the mudflats. The track-bearing horizon The track-site sediments indicate in some detail the sequence of events. The layers of mudstone beneath the footprint-bearing sandstone were presumably deposited subaqueously; then the lake dried, exposing and cracking the mud. Several closely spaced showers of rain brought sheetfloods of sand across the cohesive dry mud surface, creating ripples and laminae. The rain continued after the sand had been deposited, showering the sand surface, and pitting the tops of the ripples. Water gathered in the hollows between ripples and formed a shallow pool lapping up on to the sand. A thin mud drape fell from suspension on to the sand. The rain then stopped, and a small tetrapod swam to the edge of the pool, paddling itself on to the shore, and scratching the bottom as it did so. It pulled itself out of the water, and picked its way over the firmer ripple crests until it escaped from the immediate water’s edge. At the beginning of the track, as preserved (Text-fig. 4), the animal was apparently swimming, as shown by the incomplete scratch marks produced by the toes contacting the sediment. As it emerged from the water, the animal crossed a narrow zone of water-saturated sand, into which it sank, leaving ill- defined prints. Then, moving on to drier sand, the trackway shows that the animal walked round a water-filled depression. Of the clear prints near the top of slab 5 (Text-fig. 4) two show impressions of a slipping foot and a hand close together, presumably because the small animal had to struggle to surmount the steep ripple facing it. The track was produced after a shower of rain, as indicated by the fact that the rain imprints have not erased the footprint impressions. After the animal had passed over this narrow belt of land, the water in the ripple hollows soaked away through the sand, or evaporated in the rays of a hot sun. The whole process probably lasted little longer than 24 hours. Shortly after, the pool or lake again flooded the site, covering the footprint-bearing layer with grey mud and sand. Typical freshwater TVERDOKHLEBOV ET AL.: PERMIAN FOOTPRINTS 165 organisms, such as conchostracans and ostracods, as well as fishes and bivalves, occupied the new lake. Acknowledgements. MJB and GWS thank the Royal Society for funding their visits to Russia in the summers of 1994 and 1995, as part of the Royal Society Joint Research Programme between palaeontologists in Russia and Bristol. MJB and GWS also thank their hosts from Saratov State University, especially Professor V. G. Ochev, for their great kindness and assistance. GWS acknowledges the contribution of research facilities at the Geier Collections and Research Center, Cincinnati Museum of Natural History, and the Department of Geology, University of Cincinnati, for aiding this work. We thank Andrew Milner for helpful comments on the nature of the track-maker. REFERENCES ANFIMOV, L. V., CHUVASHOV, B. I., GRIFER, B. I., GUSEV, A. K., ROCKLIN, A. I., MURAVYEV, I. S., SOFRONITSKY, P. A. and zolotova, v. p. 1993. General characteristics of the Permian deposits of the Urals and Povolzhye: Upper Permian. In Permian System: guides to ecological excursions in the Uralian type localities. Occasional Publications , Earth Sciences and Resources Institute, University of South Carolina, Columbia, New Series, 1©, 24—33. demathieu, G., gand, G. and toutin-morin, N. 1992. La palichnofaune des bassins permiens proven<;aux. Geobios, 25, 19-54. efremov, i. a. and vjushkov, b. p. 1955. [Catalogue of localities of Permian and Triassic terrestrial vertebrates in the territories of the U.S.S.R.]. Trudy Paleontologicheskiy Instituta, 46, 1-185. [In Russian], fichter, J. 1983. Tetrapodenfahrten aus dem saarpfalzischen Rotliegenden;? Ober-Karbon, Unter-Perm; Siidwest-Deutschland, 1 : Fahrten der Gattungen Saurichnites, Limnopus, Amphisauroides, Protritonichnites, Gilmoreichnus, Hyloidichnus und Jacobiichnus. Mainzer Geowissenschaftlichen Mitteilungen , 12, 9-121. gand, g., demathieu, g. and ballestra, f. 1995. La palichnofaune de vertebres tetrapodes du Permien Superieur de l’Esterel (Provence, France). Palaeontographica, Abteilung A, 235, 97-139. garyainov, v. a., kulyova, g. v., ochev, v. G. and tverdokhlebov, v. p. 1967. [ Guidebook to excursions through Upper Permian and Triassic continental sediments of the south-west of the Russian Platform and Pre- Urals], Izdatelstvo Saratovskogo Universiteta, Saratov, 148 pp. [In Russian], haubold, h. 1971. Ichnia amphibiorum et reptiliorum fossilium. Handbuch der Paldoherpetologie , 18, 1-124. kozur, H. 1993. Boundaries and subdivisions of the Permian System. In Contributions to Eurasian Geology. Occasional Publications, Earth Sciences and Resources Institute, University of South Carolina, New Series, 9B, 139-154. molostovskaya, i. i. 1982. [History of development of Late Permian ostracods of the subfamily Darwinulacea on the Russian Platform and their significance for regional stratigraphy]. Voprosi Micropaleontologii, 25, 155-163. [In Russian], 1993. Nonmarine ostracods and paleobiogeographical distribution of Late Permian basins in the Eastern Russian Plate. In Contributions to Eurasian Geology. Occasional Publications , Earth Sciences and Resources Institute, University of South Carolina, Columbia , New Series, 9B, 95-100. Newell, a. j., tverdokhlebov, v. p. and benton, m. j. in press. Sedimentology of late orogenic (Permian, Tatarian) molasse from the southern Ural foreland basin. Sedimentology. nopcsa, f 1923. Die Familien der Reptilien. Fortschritte der Geologie und Palaontologie, 2, 1-210, pis 1-6. ochev, v. g., tverdokhlebova, G. I., minikh, m. G. and minikh, a. v. 1979. [Stratigraphical and palaeo- geographical importance of Upper Permian and Triassic vertebrates of the East European platform and Pre- Urals], Izdatelstvo Saratovskogo Universiteta, Saratov, 160 pp. [In Russian]. olson, e. c. 1957. Catalogue of localities of Permian and Triassic terrestrial vertebrates of the territories of the U.S.S.R. Journal of Geology, 65, 196-224. picard, m. d. and high, l. r., Jr 1972. Sedimentary cycles in the Green River Formation (Eocene), Uinta Basin, Utah. Journal of Sedimentary Petrology , 38, 378-383. stear, w. m. 1983. Morphological characteristics of ephemeral stream channel and overbank splay sandstone bodies in the Permian Lower Beaufort Group, Karoo Basin, South Africa. 405-420. In collinson, j. d. and lewin, j. (eds). Modern and ancient fluvial systems. International Association of Sedimentologists Special Publication, 6. Blackwells, Oxford, 527 pp. tverdokhlebov, v. p. 1988. [Some general problems of investigation of arid continental ecosystems]. Manuscript N 359-B 88, deposited in Saratov State University, Saratov. [In Russian]. 166 PALAEONTOLOGY, VOLUME 40 1989. [Methodological aspect. Genetic types of continental sediments of the arid and semi-arid zone and taphononnc peculiarities in connection with their localities of the remains of tetrapods], 66-73. In [Theory and experiment in taphonomy. Materials of the scientific seminar on questions of taphonomy and paleoecology, Saratov , May 1986], Izdatelstvo Saratovskogo Universiteta, Saratov, 162 pp. [In Russian]. tverdokhlebov, G. I. 1976. [Catalogue of tetrapod localities of the Upper Permian of the southern Pre-Urals and the south-eastern Russian Platform ]. Izdatelstvo Saratovskogo Universiteta, Saratov, 88 pp. [In Russian]. VALENTIN P. TVERDOKHLEBOV GALENA I. TVERDOKHLEBOVA Nauchno-Issledovatelskii Institut Geologii Prospekt Moskovskaya, 161 410750 Saratov, Russia MICHAEL J. BENTON Department of Geology University of Bristol Bristol BS8 1RJ, UK GLENN W. STORRS Department of Geology University of Bristol Bristol BS8 1RJ, UK and Geier Collections and Research Center Cincinnati Museum of Natural History 1720 Gilbert Avenue Cincinnati, Ohio 45202, USA Typescript received 22 November 1995 Revised typescript received 27 March 1996 LOWER CAMBRIAN CAMBROCLAVES ( INCERTAE SEDIS) FROM XINJIANG, CHINA, WITH COMMENTS ON THE MORPHOLOGICAL VARIABILITY OF SCLERITES by S. CONWAY MORRIS, J. S. CRAMPTON, XIAO BING and A. J. CHAPMAN Abstract. Four species of cambroclave, an enigmatic group whose position within the Metazoa is unresolved, are described from the Lower Cambrian Yurtus Formation of western Xinjiang, China. Cambroclavus bicornis is similar to a number of previously described species, including C. absonus from approximately equivalent age strata of South Australia. As in many cambroclave taxa, morphological variability of individual sclerites appears to be considerable. Morphometric analysis of four populations of C. bicornis, using elliptic Fourier shape analysis (EFA), demonstrates that this technique offers considerable discriminatory power. Two samples from the upper parts of one section (Sugaitbulak) show morphological stasis. They are also readily distinguishable from two other samples, which although from near-equivalent horizons in adjacent sections on Yurtus Mountain differ significantly from one another. Zhijinites claviformis is a robust sclerite, with a strongly ribbed spine. In contrast, Parazhijinites cf. guizhouensis has a remarkably slender spine, arising from a much reduced base. Finally, Wushichites minutus is more reminiscent of other Chinese cambroclaves, notably Deiradoclavus trigonus, with a sclerite with a sub-circular base indented by a posterior notch, and a much reduced spine. Although the overall morphology of cambroclaves appears to be related to protection from predatory attack, the wide variation in basic sclerite types lacks convincing ecological explanations. The most obvious manifestation of the rapid diversification of metazoans close to the Vendian-Cambrian boundary is the relatively abrupt appearance of skeletal parts. Many have been long familiar, such as the remains of trilobite exoskeletons, brachiopod valves, and echinoderm ossicles. Others were once enigmatic, but are now recognized as dispersed units of scleritomes. An example of the latter are the halkieriids, slug-like metazoans with a cataphract armour of sclerites (Conway Morris and Peel 1995). There remains, however, a considerable number of Cambrian taxa which are of very uncertain systematic position. They have no obvious affinity to any known major group, and often show at least a moderate diversity of form. The cambroclaves exemplify this problem: they are represented by a variety of sclerite morphs, typically phosphatized during diagenesis but with good evidence of an original composition of calcium carbonate. The sclerites have the common form of a basal unit and projecting spine. The interior of the sclerite is hollow, and originally was presumably filled with soft tissue and/or fluid. A somewhat crude sub-division recognizes four types of sclerites. Zhijinitids and paracarinachitids have an approximately oval base and prominent spine. Cambroclavids have a ‘dumb-bell’ shape, with the spine located at the anterior (arbitrarily defined) end. Deiradoclavids (and the similar wushichitids) are approximately circular and the spine may form a transverse ridge. Finally, deltaclavids are approximately tear- drop-shaped with the short spine located at the expanded anterior end. The great majority of sclerites are found isolated. Rare examples of fused assemblages, combined with analysis in single sclerites of congruent outlines and articulatory facets, demonstrate, however, that at least some areas of the scleritome could have formed interlocking arrays, if not sheets. Of particular significance was the independent recognition of relatively extensive fused clusters in [Palaeontology, Vol. 40, Part 1, 1997, pp. 167-189, 4 pis] © The Palaeontological Association 168 PALAEONTOLOGY, VOLUME 40 Chinese material by Conway Morris and Chen (1991) and Yue (1991). Of equal interest was the observation that the sclerites can form two layers, back-to-back. This suggests that, unless this configuration arose by some sort of post-mortem folding, which seems rather unlikely, the scleritome arrangement was not a simple cataphract armour covering the surface of a slug-like animal. A further observation is that in a fused assemblage of four parallel rows, the two central ones are notably smaller than those flanking on either side (Yue 1991, fig. 1.5). The affinities of cambroclaves remain enigmatic. Comparisons between the zhijinitids and the hooks of parasitic worms such as the Acanthocephala (e.g. Qian and Yin 1984) are considered unlikely (Conway Morris and Chen 1991). The apparent arm-like nature of some assemblages allows a very tentative comparison with the echinoderms (Conway Morris 1993), but the sclerites do not show the characteristic stereom ultrastructure of this phylum. Another possibility should also be noted. One of the priapulid worms (Cricocosmia jinningensis) from the Lower Cambrian Konservat-Lagerstatte of Chengjiang, Yunnan province, China, has a double row of sclerites that extend along almost the entire trunk (Hou and Sun 1988; Hou and Bergstrom 1994). These sclerites have received only cursory description, but examination of material in Nanjing (courtesy of Hou Xianguang and Chen Junyuan) indicates that they have a distinctive structure. Each sclerite is somewhat elliptical in outline, and concavo-convex. The upper surface is traversed by a low ridge at the anterior. The lower concave surface bears a doublure-like structure on its anterior margin. There are also a number of undescribed worms from the Chengjiang biota that carry other types of discoidal and conical sclerites. None of these is directly comparable to known cambroclaves but an affinity with the priapulids or some other group of protostome worms cannot be ruled out. Finally, it might be noted that the gastrotrichs, a phylum of microscopic animals that are usually placed in the aschelminths, sometimes have cuticular spines that have quite striking similarities to cambroclaves (e.g. Evans 1992; Balsam and Fregni 1995). Nevertheless, at the moment no other evidence appears to support a phylogenetic connection between these two groups. While a position with the Metazoa still seems more plausible, it may also be worth speculating whether cambroclaves could be some type of algae. In particular Dzik (1994, fig. 19) illustrated isolated meroms of an Ordovician receptaculitid, which resemble cambroclavid sclerites. They also have a fibrous ultrastructure that resembles that documented in cambroclaves (e.g. Bengtson et al. 1990, figs 64s, 65p, s, w, 67o), and was interpreted by Dzik (1994) as a replacement of an acicular aragonite ultrastructure. Receptaculids are widely regarded as algal, perhaps related to the dasycladeans (e.g. Nitecki 1986), but Dzik (1994) has reiterated the possibility of a sponge relationship. The history of research into cambroclaves and their palaeobiology were both reviewed at some length by Conway Morris and Chen (1991), and will not be repeated here. Since then, however, several items require discussion. First, relatively few new reports on cambroclaves have been issued. An important record, however, is from sections in Shaanxi province that have a number of faunal similarities to the equivalent-aged strata in Xinjiang (Ding et al. 1991). The taxonomy of the Shaanxi cambroclaves requires some revision, but in particular the material ascribed to Zhijinites sp. (Ding et al. 1991, pi. 2, figs 29-31, 36) from sections near Xixiang approaches quite closely the material described herein as Zhijinites claviformis. Another new report is from the upper Atdabanian of Gorlitz, eastern Germany where Elicki (1994, fig. 6.17-6.20; see also Elicki and Schneider 1992, pi. 16, figs 10-1 1 ; Geyer and Elicki 1995) documents Cambroclavus ludwigsdorfensis. More information is also available on the cambroclaves from the Yurtus Formation of western Xinjiang (e.g. Duan and Xiao 1992; Xiao and Duan 1992; Yue and Gao 1992), but as these form the main point of this paper these reports are assessed in more detail below. Considerably more doubt surrounds fossils, from the Lower Cambrian of Cape Breton Island, that Landing (1991, p. 591) described as Samsanoffodavus matthewi and proposed tentatively might be 'a zhijinitiid or zhijinitiid relative’. 5. matthewi consists of hollow sclerites, with a pointed apical region extending into a much broader base, which defines a wide aperture. Although zhijinitids have a spine and base, the demarcation is generally much stronger and the base is usually closed. Another area of confusion that remains unresolved is the appropriate taxonomy for CONWAY MORRIS ET AL.: LOWER CAMBRIAN CAMBROCLAVES 169 70° 80° 90° 100° 110° 120° 140° 79°02' 80°30' text-fig. 1 Locality maps, a, central Asia showing the position of the town of Aksu. Urumqi is the capital of Xinjiang. Alma Ata is in Kazakhstan. The localities that yield the cambroclaves in this region (see text) are situated to the east of Alina Ata. b, detailed locality map of the Aksu-Wushi area, showing the position of the six stratigraphical sections that yielded cambroclaves. 170 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Stratigraphical distribution of the four cambroclave taxa in six sections of the Yurtus Formation, Aksu-Wushi, Xinjiang, China. The underlying unit is the Qigebulak Formation, the overlying unit is the Xiaoerbulak Formation. The numbers 9-10, 16-18, 31-35, 37-38, 42-44 and 47 refer to field numbers of samples collected in 1991. Other samples were collected earlier by one of us (XB): 5/B, 5/D-5/F are from Sugaitbulak; XXIV-7/1, 7/2 and 7/10 are from Section VIII. See Text-figure 1 for geographical locations of these sections. Stratigraphical sections are based on Figure 2.6 of Gao et al. (1985a). CONWAY MORRIS ET AL.: LOWER CAMBRIAN CAMBROCLAVES 171 cambroclaves, especially at the generic and specific levels. Despite discussions of morphological variability (see also below), which in at least some populations is marked (Bengtson et al. 1990; Conway Morris and Chen 1991), and the reiterated need to consider larger sample sizes than is current practice, more recent reports tend to document material or even erect new taxa on limited appraisal of scleritome variability (e.g. Xiao and Duan 1992; Yue and Gao 1992; Elicki 1994). In addition, the status of a number of cambroclave genera is unresolved. Duan (1984) erected Phyllochiton, Sinoclavus and Tanbaoites for cambroclave material recovered from the Xihaoping Formation of Shennongjia, Hubei province. He also described a number of new species of Cambroclavus. Duan’s (1984) actions were reviewed by Bengtson et al. (1990, p. 103) who tentatively proposed that all three genera were junior synonyms of Cambroclavus. For the most part, the nominal species of cambroclave from the four sections documented in Shennongjia by Duan (1984) overlap (see his Table 1), and they may all derive from a single species. Yue (1991), however, continued to recognize Phyllochiton , following Duan’s (1984) discrimination of sclerites bearing lateral spurs. Such structures, however, are variably present in species of Cambroclavus , such as C. absonus from South Australia (Bengtson et al. 1990, figs 64a-b, r, 65g, t) and C. bicornis from Xinjiang (see below). There is, however, a danger of over-enthusiastic forays into synonymy. In particular, a distinctive type of cambroclave, with a small base and a very elongate, narrow spine, is referred to as Parazhijinites guizhouensis (Qian and Yin 1984, pi. 2, figs 1-8). It was first described from Guizhou province, and subsequently was tentatively synonymized with Zhijinites longistriatus by Conway Morris and Chen (1991, p. 367). In this paper, however, it is now restored as a separate taxon, with possible new records from western Xinjiang (see below). Finally, it should be noted that the palaeobiogeography of the cambroclaves has received little attention, although Jiang (1992, fig. 8) invoked a diachronous migration across central and south Asia. The purpose of this paper is two-fold. First, it redescribes and reassesses the cambroclave fauna of Xinjiang with detailed descriptions of Cambroclavus bicornis (Qian and Xiao, 1984), Zhijinites claviformis Qian and Xiao, 1984, Parazhijinites cf. guizhouesis Qian and Yin, 1984, and Wushichites minutus Qian and Xiao, 1984. Secondly, a quantitative consideration of cambroclave variability is provided, using elliptic Fourier shape analysis (EFA) of outlines from selected populations of C. bicornis. STRATIGRAPHY AND LOCALITIES The geographical locations of the studied sections in the Yurtus Formation are shown in Text- figure 1 . A summary of the Vendian-Cambrian stratigraphy of this area and the sections sampled is given by Conway Morris and Chapman (in press), and only a few comments will be given here. The Yurtus Formation (Text-fig. 2) is of variable thickness, and consists of a mixture of calcareous elastics and impure limestone, although the basal unit is a siliceous phosphorite that has been mined quite extensively. Text-figure 2 depicts the distribution of the cambroclave taxa, and further comments for each species are given below. In general, occurrences tend to be concentrated near the top of the Yurtus Formation, where there is evidence for sedimentary condensation and phosphatization, prior to the abrupt transition to the ribbon-bedded limestones of the overlying Xiaoerbulak Formation (from which one sample was obtained). The abundance of cambroclaves in these horizons is, therefore, mostly taphonomically enhanced, and the occurrence of Cambroclavus bicornis low in the section at Aksu phosphorite mine (Text-figure 2) shows that this species of cambroclave at least has a relatively long range. Of the six sections that yielded one or more species of cambroclave, five were visited in 1991. The sixth, at Sugaitbulak, was inaccessible owing to deterioration of the track, and the productive samples were obtained earlier by one of us (XB). The age of the Yurtus Formation and its inferred correlation with other Precambrian-Cambrian units is not entirely resolved. Recovery of Rhombocorniculum cancellation from the Yurtus Formation (Yue and Gao 1992) indicates, however, that the bulk of this unit is Atdabanian. The most secure correlations appear to exist with the sequences in Kazakhstan, especially the Maly 172 PALAEONTOLOGY, VOLUME 40 Karatau region, on account of the general faunal similarities. Much of the Yurtus Formation appears to correlate with Chulaktau Formation and the lower part of the overlying Shabakty Formation. These correlations and the general faunal aspect of the Yurtus Formation also support this age assignment. A more detailed discussion of these topics is given by Conway Morris and Chapman (1996, in press). SYSTEMATIC PALAEONTOLOGY Class cambroclavida Conway Morris and Chen, 1991 Family zhijinitidae Qian, 1978 Genus cambroclavus Mambetov in Mambetov and Repina, 1979 Cambroelavus bicornis (Qian and Xiao, 1984) Plate 1 ; Plate 2, figures 1-9 1984 Sugaites bicornis Qian and Xiao, p. 79, pi. 1, figs 9, 12-13; pi. 3, figs 16-19. 1984 Sugaites soleiformis Qian and Xiao, p. 80, pi. 1, figs 10-11; pi. 3, figs 14—15. 1984 Sugaites sicyojdeus Qian and Xiao, p. 80, pi. 3, fig. 23. 1984 Sugaites hastatus Qian and Xiao, p. 81, pi. 3, figs 20-22. 1984 Sugaites saccatus Qian and Xiao, p. 81, pi. 3, figs 24-26. 1984 Cambroclavus paradoxus Qian and Yin, p. 220, pi. 2, figs 9-12. 1985a Sugaites bicornis', Gao et ah, pi. 5, figs 9-11; pi. 8, figs 9-11. 1985a Sugaites soleiformis ; Gao et al., pi. 5, figs 12-13; pi. 8, figs 12-14. 1985a Sugaites saccatus', Gao et al., pi. 8, figs 6-8. 1985a Sugaites hastatus', Gao et al., pi. 8, figs 15-18. 19856 Sugaites bicornis', Gao et al., pi. 34, fig. 21. 19856 Sugaites soleiformis', Gao et al., pi. 34, fig. 22. 1985 Cambroclavus bicornis ; Wang et al., pi. 3, fig. 13. 1992 Cambroclavus sp. ; Yue and Gao, pi. 4, fig. 4. 1992 Sinoclavus Xiao and Duan, p. 225, pi. 3, figs 24—25. 1992 Cambroclavus bicornic [«c]; Xiao and Duan, pi. 4, fig. 35. 1992 Cambroclavus spp.; Xiao and Duan, pi. 4, figs 27, 32, 37. 1992 Cambroclavus soleiformis', Xiao and Duan, pi. 4, figs 14-15. 1992 Cambroclavus hastatus', Xiao and Duan, pi. 4, figs 13, 24—26, 30. 1992 Phyllochiton vescus [nomen nudum]', Xiao and Duan, pi. 4, figs 3-4. 1992 Phyllochiton tubercularis [nomen nudum] ', Xiao and Duan, p. 225, pi. 4, figs 1-2, 8. 1992 Phyllochiton tubercularis ; Duan and Xiao, p. 343, pi. 1, figs 17-18. 1992 Phyllochiton vescus', Duan and Xiao, p. 343, pi. 2, figs 17-18. 1992 Cambroclavus hastatus', Duan and Xiao, pi. 3, figs 1, 15. 1992 Cambroclavus soleiformis', Duan and Xiao, pi. 3, fig. 2. 1992 Cambroclavus sp.; Duan and Xiao, pi. 3, fig. 20. 1992 Sinoclavus', Duan and Xiao, pi. 3, fig. 21. 1992 Cambroclavis bicornis ; Jiang, figs 8, 8a, 13. 1992 Cambroclavus soleiformis ; Jiang, figs 5, 7, 7a, 12. 71994 Zhijinites; Yue and Gao, pi. 12, fig. 11. Holotype. Xinjiang Institute of Geology 10583. EXPLANATION OF PLATE 1 Figs 1-12. Cambroclavus bicornis (Qian and Xiao, 1984). 1, SM X. 26167. 2, SM X. 26168. 3, SM X.26169. 4, SM X. 26170. 5, SM X.26171. 6, SM X.26172. 7, SM X.26173. 8, SM X.26174. 9, SM X.26175. 10, SM X. 26176. 11, SM X. 26177. 12, SM X. 26178. All isolated sclerites, dorsal (1-2, 4, 8-12) and ventral surfaces (3, 5-7), from the Yurtus Formation, Yurtus Mountain; section 1 (sample XI/91/31), 1-3; section 2 (sample XI/91/35), 4-12, Aksu-Wushi, Xinjiang, China. Magnifications are x 100 (1-2), x 110 (3), x 150 (4—6, 9, 11), x 175 (7-8), x 200 (10, 12). PLATE 1 CONWAY MORRIS et al., Cambroclavus 174 PALAEONTOLOGY, VOLUME 40 Material illustrated here. Sedgwick Museum (University of Cambridge) SM X.26167-X.26187. Stratigraphical horizon. Yurtus Formation (and basal-most Xiaoerbulak Formation), Atdabanian Stage, Lower Cambrian. Localities and sections. Mountains south-west of Aksu, western Xinjian, China. C. bicornis has been recovered from all sections, except for Section VIII. As Text-figure 2 shows, in some samples it co-occurs with one or more of the other cambroclave taxa, but in other instances it is the only taxon to be recovered. Preservation. The phosphatic composition of the sclerites is evidently diagenetic, after an originally calcareous composition (Bengtson et al. 1990; Conway Morris and Chen 1991). In most of the Xinjiang specimens the fidelity of overgrowth is moderately good, although the spine is usually incomplete. In a few cases original ultrastructure appears to be preserved, most notably as radiating fibres in the anterior section (PI. 2, figs 1-2). Similar ultrastructure has also been observed in cambroclaves from Australia (Bengtson et al. 1990, figs 65s, 67o, 69p). Diagnosis. Sclerites variable in shape, but most are ‘dumb-bell’ in outline with waist-like mid- region separating expanded anterior section and a posterior area of variable configuration that ranges from relatively expanded to tapering. Anterior spine relatively stout. Posterior of some sclerites bears prong-like extensions. Concave facet-like structures variably developed on lateral edges of mid-region, and in some sclerites also at posterior. Description. As has been documented elsewhere, populations of cambroclave sclerites show considerable morphological variation (e.g. Bengtson et al. 1990; Conway Morris and Chen 1991). This is also the case for C. bicornis (e.g. cf. PI. 2, figs 3 and 6), although overall many sclerites approximate to a dumb-bell-like shape. In general the sclerites are approximately symmetrical (e.g. PL 1, figs 1-2, 7; PI. 2, figs 1, 3, 6-8), but some sclerites show a considerable degree of asymmetry (e.g. PI. 1, fig. 8; PI. 2, fig. 9). The great majority of sclerites consist of an anterior region, more or less circular in outline, that extends posteriorly as a broad shaft. Very often this posterior region is also expanded in width. Variations on this arrangement are common. In some specimens (e.g. PI. 2, fig. 3) the posterior region shows a more pronounced taper, whereas in others (e.g. PI. 1, fig. 9; PI. 2, fig. 9) there is little change in width towards the posterior. In at least one sclerite (PI. 1, fig. 12) the posterior is strongly truncated, so that it approaches a zhijinitid-like condition. This does not appear to be due to breakage, and a similar condition is more common in Cambroclavus absonus from the Lower Cambrian of South Australia (Bengtson et al. 1990). The anterior spine is generally truncated (e.g. PI. 1, figs 4, 8-9; PI. 2, figs 4, 9), but this is evidently a result of incomplete fossilization and in well-preserved sclerites the spine is stoutly conical (PI. 1, fig- 1 1 ; PI- 2, fig. 5). The mid-section of the sclerite tends to consist of a central ridge with flanking edges, that form variably defined arcuate embayments (e.g. PI. 1, fig. 10; PI. 2, figs 2, 4, 7-8) whose original function is inferred to have been articulatory facets to accommodate the expanded ends of adjacent sclerites (see Bengtson et al. 1990, fig. 70; Conway Morris and Chen 1991, text-fig. 11). Occasionally, the central ridge has a lobe-like development (PI. 1, fig. 4). The central region of the posterior end is sometimes excavated to form a shallow facet (e.g. PI. 1, figs 4, 9; PI. 2, figs 4, 8). As with the lateral facets this embayment presumably housed the arcuate lower surface of the next-posterior sclerite in the longitudinal file. In some specimens this facet is also associated with a variably developed notch which thereby defines lobe-like extensions (PL 1, figs 4, 6, 10; PL 2, fig. 8). In one sample (XI /9 1/31; section 1, Yurtus Mountain; Text-fig. 2) the population has a conspicuous number of EXPLANATION OF PLATE 2 Figs 1 -9. Cambroclavus bicornis (Qian and Xiao, 1984). 1, SM X. 26179. 2, SM X. 26180. 3, SM X.2618 1 4, SM X. 26182. 5, SM X.26183. 6, SM X.26184. 7, SM X.26185. 8. SM X.26186. 9, SM X.26187. Figs 10-12. Zhijinites claviformis (Qian and Xiao. 1984). 10, SM X.26188. 1 1, SM X.26189. 12, SM X.26190. All isolated sclerites, dorsal (1-9) and lateral (10-12) surfaces, from the Yurtus Formation, Yurtus Mountain section 2 (sample XI/91/35), 1-9; section 6 (sample XXIV-7/2), 10-12, Aksu Wushi, Xinjiang, China. Magnifications are x 150 (1-4, 9), x 200 (4), x 175 (5-6, 8, 10-11), x 125 (12). PLATE 2 CONWAY MORRIS et ai, Cambroclavus, Zhijinites 176 PALAEONTOLOGY, VOLUME 40 sclerites with the posterior region bearing a pair of rather prominent prong-like extensions (PI. 1, figs 1-3). The significance of the morphometric variation in C. bicornis is discussed below. Remarks. The synonymy list given above, where a number of species is subsumed within C. bicornis , is based on the reasonable assumption that, as in many other cambroclaves, the sclerites of this species show considerable morphological variation and more importantly cannot be easily divided into species. Many of the features described here have already been illustrated by Chinese workers, albeit without detailed comment. Thus, features such as prominent posterior prongs (Xiao and Duan 1992, pi. 4, figs 1-4) and posterior notch (Qian and Xiao 1984, pi. 3, figs 16, 18-19) are already known, but used for interspecific distinctions that we prefer to interpret as variability within a single species (but see below). Comparisons with previous descriptions of cambroclaves are for the most part difficult because few workers have adopted scleritome-based interpretations. It is evident, however, that close similarities exist between C. bicornis and C. absonus from the Lower Cambrian of Australia. While synonymy remains possible, the Australian taxa appears to differ in terms of the relative abundance of zhijinitid-like morphs (Bengtson et al. 1990, figs 69b-c, e-f, h-p). In addition, although the morphological variability of both species is very wide and counterparts between nearly all types of sclerite shape can be found, in C. absonus there seems to be a greater tendency towards sclerites with a flared posterior section (Bengtson et al. 1990, figs 64o, x; 65a, t-u, 65x, 67d), whereas in C. bicornis such morphs are rare (Xiao and Duan 1992, pi. 4, fig. 37). In general, the strongest faunal similarities of the Xinjiang assemblages are with Kazakhstan and to a lesser extent South China, as is evident for example from a comparison of the halkieriids (Conway Morris and Chapman in press). In the case of the cambroclaves, however, matters are somewhat more complicated. As documented below in terms of zhijinitid morphs, there appears to be a significant similarity between Kazakhstan and Xinjiang faunas in the form of Zhijinites undulatus and Z. claviformis respectively. Unfortunately, the type species of Cambroclavus , C. antis from Kazakhstan, is relatively poorly documented, although it is known from articulated arrays (Mambetov and Repina 1979). C. antis is broadly similar to C. bicornis , but appears to differ in an accentuated anterior section with characteristic radial furrows on its underside, and relatively pronounced longitudinal ridges on upper and lower surfaces. In terms of the Lower Cambrian sequences on the South China platform, the most extensive documentation of Cambroclavus is by Qian and Zhang (1983) and Duan (1984), both of whom illustrated material from Hubei Province. The latter publication erects what we regard as a plethora of form-taxa. In apparent contrast, Qian and Zhang (1983) place all their material in a single species (C. fangxianensis), but the morphological variability of the sclerites they illustrate is considerably less. Overall, the similarities between C. bicornis and the material from Hubei do not seem to be very pronounced. There is, however, an indication of a stronger resemblance between material placed in various species of Sinoclavus (Duan 1984, pi. 5, figs 11-15) and C. antis from Kazakhstan, most notably in the shared possession of the ridges radiating across the lower surface of the anterior unit. Accompanying the sclerites of Cambroclavus in the Hubei samples is a distinctive deiradoclavid form that Qian and Zhang (1983) assign to Isoclavus, and Duan (1984) placed in Tanbaoites. These EXPLANATION OF PLATE 3 Figs 1-19. Parazhijinites guizhouensis Qian and Yin, 1984. 1, SM X. 26191. 2, SM X. 26192. 3, SM X. 262193. 4, SM X. 26194. 5, SM X.26195. 6, SM X. 26196. 7, SM X.26197. 8, SM X.26198. 9, SM X.26199. 10, SM X. 26200. 11, SM X.26201. 12, SM X.26202. 13, SM X. 26203. 14, SM X. 26204. 15, SM X.26205. 16, SM X. 26206. 17, SM X.26207. 18, SM X.26208. 19, SM X.26209. All isolated sclerites from the Yurtus Formation, Aksu phosphorite mine; section 3 (sample XI/91/17), 1-6, 8, 10, 15-16; (sample XI /9 1/16), 7, 9, 1 1, 17-19; section 6 (sample XXIV-7/2), 12-14; Aksu-Wushi, Xinjiang, China. Magnifications are x 125 (1,4, 6-7, 13, 18), x 175 (2, 10), x 150 (3, 5, 19), x 200 (8, 11, 14-15, 17), xl00 (9), x 225 (12), x 250 (16). PLATE 3 CONWAY MORRIS et al., Parazhijinites 178 PALAEONTOLOGY, VOLUME 40 latter two genera are evidently synonymous. The sclerites are sub-circular in outline, apart from a rather prominent posterior notch. The spine is relatively robust, and tends to be transversely elongated. Bengtson et al. (1990, pi. 115) proposed that these sclerites may derive from the same scleritome as C. fangxianensis. Wushichites , described below, is quite similar to Isoclavus, but its distribution in the Xinjiang samples appears to be more consistent with it forming a separate taxon. Zhijinites claviformis Qian and Xiao, 1984 Plate 2, figures 10-12 1984 Zhijinites claviformis Qian and Xiao, pi. 81, pi. 2, fig. 14; pi. 3, fig. 27. 1985a Zhijinites claviformis', Gao et al., pi. 5, fig. 14; pi. 8, figs 1-3. 1985/) Zhijinites claviformis', Gao et al., pi. 34, fig. 20. 1985 Zhijinites claviformis', Wang et al., pi. 3, figs 11-12. 71991 Zhijinites sp.; Ding et al., p. 103, pi. 2, figs 29-31, 36. 71992 Zhijinites longistriatus; Yue and Gao, pi. 3, fig. 8. 1992 Zhijinites deltatus', Xiao and Duan, p. 225, pi. 3, figs 17, 19-20. 1992 Zhijinites clavus', Xiao and Duan, p. 224, pi. 3, fig. 18. 1992 Zhijinites planispinosus', Xiao and Duan, p. 225, pi. 3, fig. 23. Holotype. Xinjiang Institute of Geology 03109. Material illustrated here. SM X.26188-X.26190. Stratigraphical horizon. Yurtus Formation, Atdabanian Stage, Lower Cambrian. Localities and sections. This species occurs in all sections, except for section I on Yurtus Mountain, but it is found in fewer of the sampled horizons than C. bicornis (Text-fig. 2). Diagnosis. Robust zhijinitid morph. Base sometimes quite steep with prominent radial and sometimes annular ornamentation. Stout spine eccentrically located, towards presumed anterior, with prominent longitudinal ribbing. Description. In contrast with most cambroclaves, the sclerites consist of a relatively small base and a massive spine. The spine is located towards the presumed anterior end of the base. The base typically bears pronounced, somewhat irregular, radial ornamentation. In addition, there may be annular structures that give the base a stepped appearance (PI. 2, fig. 1 1). For the size of the sclerite the spine is remarkably robust, when compared with other zhijinitids. It bears prominent longitudinal ribbing. Remarks. This species has been illustrated by a number of Chinese workers (e.g. Qian and Xiao 1984; Xiao and Duan 1992), but as the synonymy list indicates there seems to be little justification for the recognition of so many nominal species. In terms of other cambroclaves, Z. claviformis appears to approach most closely Z. undulatus from Kazakhstan (Mambetov and Repina 1979, p. 123, pi. 13, figs 1, 4, 6, 11-12; note that there is some evidence that the specimen illustrated in pi. 13, figs 2, 10, 13 is distinct and may derive from the co-occurring C. antis, see Bengtson et al. 1990, p. 1 13). The principal points of difference between Z. claviformis and Z. undulatus appear to be that in the latter taxon the base tends to be more raised and the spine, while robust and strongly ribbed, tapers much more abruptly. As Bengtson et al. (1990) also noted in discussing the Kazakhstan cambroclaves there is reasonable evidence that Z. undulatus derives from a different scleritome than C. antis. Of the 16 horizons in the Yurtus Formation that have yielded C. bicornis , only in four does it co-occur with Z. claviformis, and the latter taxon occurs in three other horizons (Text-figure 2). This supports the likelihood that Z. claviformis derives from a scleritome separate from C. bicornis. CONWAY MORRIS ET AL.: LOWER CAMBRIAN CAMBROCLAVES 179 Parazhijinites cf. guizhouensis Qian and Yin, 1984 Plate 3 1984 Parazhijinites quizhouensis [.s/c] Qian and Yin, p. 220, text-fig. 3.3; pi. 2, figs 1-8. 1984 Parazhijinites quizhouensis [.v/c] ; Wang et al., p. 177, pi. 22, figs 5-8. 71984 Zhijinites intermedins; Qian and Xiao, p. 82, pi. 3, figs 28-29. 71985 Zhijinites intermedins; Gao et al., pi. 8, figs 4^5. Holotype. Nanjing Institute of Geology and Palaeontology 68238 (Qian and Yin 1984, pi. 2, fig. 2). Material illustrated here. SM X.26190-X. 26208. Stratigraphical horizon. Yurtus Formation, Atdabanian Stage, Lower Cambrian. Also known from the Gezhongwu member of the Niutitang Formation, Guizhou Province (see Wang et al. 1984). Localities and sections. Mountains south-west of Aksu, western Xinjiang, China. P. cf. guizhouensis has been recovered from two adjacent samples in the mid-section of Aksu Phosphorite Mine and from a sample close to the top of the Yurtus Formation in Section VIII (Text-fig. 2). At the Aksu Phosphorite Mine P. cf. guizhouensis does not occur with any other taxon of cambroclave, whereas in Section VIII it co-occurs with Zhijinites claviformis. Diagnosis. Zhijinitid sclerite with small base, sometimes very reduced. Radial ornamentation on base, reduced on smaller bases to more nodular appearance. Spine exceptionally elongate, circular to transversely compressed. Subdued longitudinal ribbing, occasionally central ridge on anterior side. Description. The most noticeable feature of this species is the elongate spine arising from a small base. The degree of reduction of the base is quite variable, ranging from a relatively elongate posterior section (PI. 3, figs 7, 17-19) to ones that are reduced to little more than the diameter of spine (PI. 3, figs 8, 14). On the larger bases the ornamentation consists of a subdued structure, whereas in the reduced bases the ornamentation is more nodular (e.g. PI. 3, figs 8, 12-13). In comparison with most other cambroclaves the spine is exceptionally elongate, and its tip is almost invariably missing. Where the base is fairly well developed the angle between it and the spine is relatively oblique, at about 55°. Where the base is reduced, however, the spine is almost vertical. The cross sectional shape of the spine also appears to be related to the size of the spme relative to its base. Thus, in sclerites with a reduced base the spine has a circular cross section, whereas when the base is better developed the spine is more flattened in the transverse plane. Many spines bear a subdued longitudinal ribbing, but in the more compressed spines the anterior side may bear a central ridge (PI. 3, figs 11, 18). Remarks. In an earlier discussion of cambroclaves from the South China platform it was tentatively proposed that Parazhijinites guizhouensis was synonymous with Zhijinites longistriatus (Conway Morris and Chen, 1991, p. 366). The distinctiveness of comparable material from Xinjiang and the likelihood that these sclerites are not associated with any of the other cambroclave taxa indicate that P. guizhouensis is best treated as a distinctive and separate species. The poor quality of many earlier illustrations and the lack of reinvestigations still leave a number of uncertainties concerning the taxonomy of Chinese cambroclaves. One taxon that is somewhat similar to P. guizhouensis is Z. lubricus (see Conway Morris and Chen 1991 for the latter’s tentative synonymy with Z. longistriatus). In terms of other cambroclave taxa the material from Xinjiang resembles at least some sclerites of C. clavus from Kazakhstan (Mambetov and Repina 1979, p. 122, pi. 13, figs 3, 5, 7 9). This taxon remains rather poorly known, but although Bengtson et al. (1990, pp. 37, 112) questioned its place in the cambroclaves, drawing attention to fossils from the Lower Cambrian of South Australia which they referred to as Spicule A, it now seems likely that at least some of the material from Kazakhstan is genuinely cambroclave, but should be referred to as Zhijinites clavus. PALAEONTOLOGY, VOLUME 40 180 Wushichites minutus Qian and Xiao, 1984 Plate 4 1984 Wushichites minutus Qian and Xiao, p. 76, pi. 1, fig. 7; pi. 3, fig. 11. 1984 Wushichites polyedrus Qian and Xiao, p. 77, pi. 3, figs 12-13. 1985a Wushichites minutus ; Gao et al., pi. 8, figs 19-20. 1985a Wushichites polyedrus', Gao et al., pi. 8, figs 21-22. 19856 Wushichites minutus', Gao et al., pi. 34, fig. 17. 19856 Wushichites polyedrus', Gao et al., pi. 34, fig. 18. 1985 Wushichites minutus', Wang et al., pi. 3, fig. 25. 1985 Wushichites polyedrus', Wang et al., pi. 3, fig. 26. 1989 Isoclavus minutus', Qian, p. 236, pi. 61, figs 4-10. 1992 Wushichites minutus', Duan and Xiao, pi. 1, fig. 13. 1992 Wushichites minutus', Jiang, figs 5, 9-10. 1992 Wushichites minutus', Xiao and Duan, pi. 1, figs 27-28. Holotype. Xinjiang Institute of Geology 10455. Material illustrated here. SM X.26209-X. 26220. Stratigraphical horizon. Yurtus Formation, Atdabanian Stage, Lower Cambrian. Localities and sections. Wushichites minutus is relatively uncommon in the Xinjiang sections, but occurs in section 2 of Yurtus Mountain, Aksu Phosphorite Mine, Xiaoerbulak Phosphorite Mine, and the Sugaitbulak section (Text-fig. 2). It often co-occurs with C. bicornis and Z. claviformis. Diagnosis. Sub-circular sclerites, more or less bilaterally symmetrical. Anterior region may bear a subdued dome and in at least some sclerites a narrow cleft. Posterior almost invariably with prominent notch. Upper surface with radial ornamentation. Description. As noted above the sclerites of W. minutus are notable, with few exceptions (PI. 4, figs 1 1—13), for a rather thick overgrowth of phosphate that tends to obscure many of the finer details. In outline the sclerites are almost bilaterally symmetrical, and vary from sub-circular to oval. A posterior notch is almost invariably present, but it may be only slightly developed (PI. 4, figs 5-6, 9) or be variably prominent, sometimes to the extent of defining distinct lobes (PI. 4, fig. 8). The area equivalent to the anterior spine of other cambroclaves is represented by a low dome (PI. 4, figs 3, 7), and there is no clear evidence that a discrete spine was present. The phosphatic coating normally obscures the surface details. In well-preserved specimens, however, the ornamentation of the upper surface is seen to consist of fine radial striae on the upper surface (PI. 4, fig. 12). Remarks. Wushichites is most similar to Isoclavus (and the synonymous Tanbaoites), described from Lower Cambrian sections in Hubei province (Qian and Zhang 1983; Duan 1984; Qian 1989). In this latter taxon there is also a variably developed posterior notch (Qian and Zhang 1983, pi. 3, figs 9, 11-16), but the spine is prominent. Qian (1989) synonymized Wushichites and Isoclavus. The apparently more prominent spine in the latter genus is our criterion for keeping these two genera separate. While the absence of a spine in many specimens of Wushichites could be diagenetic EXPLANATION OF PLATE 4 Figs 1-13. Wushichites minutus Qian and Xiao 1984. 1, SM X. 26210. 2, SM X.26211. 3, SM X.2612. 4, SM X.26213. 5, SM X.26214. 6, SM X.26215. 7, SM X.26216. 8, SM X.26217. 9, SM X.26218. 10, SM X.26219. 11 12, SM X. 26220. 13, SM X. 26221. All isolated sclerites from the Yurtus Formation, Yurtus Mountain; section 1 (sample 81K2 H'-V-6/l), 1-9; Yurtus Mountain; section 2 (sample XI/91/34), 11-13; (sample XI/91/35), 10; Aksu-Wushi, Xinjiang, China. Magnifications are x 110 (1), x 125 (2, 6, 9, 11, 17), x 100 (3-5, 7-8), x 150 (10), x 250 (12). PLATE 4 CONWAY MORRIS et al., Wushichites 182 PALAEONTOLOGY, VOLUME 40 and linked to the overcoat of phosphate, in well preserved specimens (PI. 4, figs 11-13) the spine is not apparent. Wushichites is somewhat less similar to Deiradoclavus, the latter genus having a well-defined triradiate ridge on its upper surface. In addition, the spine is transversely elongate, and overall the sclerites of the latter appear to show a wide degree of morphological variation. The possibility that Isoclavus actually derives from the same scleritome as co-occurring cambroclaves (see above) raises the same possibility for Wushichites minutus and Cambroclavus bicornis. Of the four horizons that have yielded Wushichites , in three of them there occurs also C. bicornis. This latter genus, however, is found at substantially more horizons without Wushichites (Text-fig. 2). On balance, therefore, it seems preferable to keep these two taxa separate. SHAPE ANALYSIS OF CAMBROCLAVES The morphological variability of cambroclaves is pronounced (e.g. Bengtson et al. 1990; Conway Morris and Chen 1991). With so little known about the original nature of the scleritome, one avenue of enquiry is to document the degree of variability in terms of morphospace occupation by the sclerites. More specifically, the aim of this section is to see the manner in which populations of sclerites from particular samples in the Yurtus Formation occupy morphospace. Do such populations occupy discrete ‘clouds’ within morphospace? If so, does such a pattern represent a species with polymorphism in terms of sclerite types, or alternatively could it indicate separate taxa? Outline shape and size of the cambroclave sclerites were examined using a technique known as elliptic Fourier shape analysis (EFA). EFA, first described by Kuhl and Giardina (1982), is a biometric technique particularly suited to the description of fossils lacking many homologous landmarks (Crampton 1995). The method mathematically ‘decomposes’ a digitized, two dimensional outline into a series of harmonically related trigonometric (sine and cosine) curves. For any harmonic these curves define an ellipse in the x—y plane, and successive harmonics describe progressively smaller features of the outline. The size, shape and orientation of each harmonic ellipse are represented using four Fourier coefficients. These coefficients were computed using a modified version of program EFA written by F. J. Rohlf and S. Ferson (1985-1991 ; kindly made available by F. J. Rohlf), and modified by one of us (JSC, in association with A. Buckley and J. Sloan). An earlier, unmodified version of this program is available in Rohlf and Bookstein (1990). Coefficients from many outlines are compared using standard multivariate statistical techniques. Computational details of EFA are given in Kuhl and Giardina (1982) and Ferson et al. (1985), and some of the methodological considerations are discussed in Crampton (1995). Hitherto, EFA has been applied quite widely, including the study of bivalves (Ferson et al. 1985; Crampton in press), insect wings (Rohlf and Archie 1984) and plant leaves (White et al. 1988). Methods In the present study, outlines were digitized using a video camera attached to a binocular microscope and the image analysis software VIDS-IV (Synoptics Cambridge, formerly Ai, Cambridge). Specimens were orientated in a standard fashion, upperside visible with the spine towards the top. The sclerites were then traced in a clockwise sense from the spine, using a hand- held cursor. Previous experience indicates that errors introduced during hand digitization are negligible when compared with the morphological variation between specimens. In any EFA it is necessary to estimate the number of harmonics required to describe an outline with a given degree of precision; this can be done in a number of ways (see Crampton 1995). In the case of the cambroclaves 99-9 per cent, of the variance or ‘shape information’ is captured by the first nine harmonics. Exploratory statistical analyses revealed, however, that harmonics seven to nine contribute little to the discrimination of groups, and hence our subsequent interpretation was based on the first six harmonics. Using parameters of the first harmonic it is possible to make several normalizations to the outlines during computation of the Fourier coefficients (Ferson et al. 1985). The sclerite data were CONWAY MORRIS ET AL.\ LOWER CAMBRIAN CAMBROCLAVES 183 normalized for starting position and orientation of the outline trace for two reasons. First, EFA is rather sensitive to variations in these factors. Secondly, it was difficult to define exactly homologous orientations and starting positions during digitization. In addition, it is possible to normalize outlines for size. In a study of shape per se it is clearly desirable to eliminate the effects of size, even though we acknowledge that this information could carry some biological significance. Hence in this study shape data were normalized for size, and this information was extracted and examined separately. Size itself was examined using univariate analysis of the first harmonic amplitude, based on non- normalized data. For any outline the first harmonic represents the ‘best-fitting’ ellipse, and the amplitude of this harmonic can be taken as a proxy for outline size (Text-fig. 3). The amplitude is text-fig. 3. Three representative out- lines of cambroclave sclerites, showing first harmonic (‘best-fitting’) ellipses generated during EFA. The size, or amplitude, of these ellipses can be taken as a proxy for outline size. given by the square root of the sum of the squared Fourier coefficients. Amplitude data were compared both graphically and in an analysis of variance (ANOVA). The ANOVA was performed using the pc-based software SYSTAT v. 5.0. Shape was examined using multivariate analyses of Fourier coefficients for harmonics two to six, based on size-normalized data. Arguments for the elimination of the first harmonic in an analysis of this sort were discussed in Crainpton ( 1995). Coefficients for many outlines were interpreted using principal components (PCA) and canonical variates (CVA) analyses calculated using the pc-based program NTSYS-pc v. 1.80 (Rohlf 1993). Four samples ( 1-4) of cambroclaves were chosen for this study. The details of their location are as follows: 1, Sample 5/B from Bed 4 of the Sugaitbulak section (91 specimens; not illustrated here, but practically indistinguishable from 3 (sample 5/E), see below); 2, Sample XI/91/34 from Bed 7 of section 2 on Yurtus Mountain (105 specimens; see Text-fig. 3, left); 3, Sample 5/E from Bed 8 of the Sugaitbulak section (250 specimens; see Text-fig. 3, right); and 4, Sample XI/91/3 from Bed 3 of section 1 on Yurtus Mountain (45 specimens; see PI. 1, figs 1-3 and Text-fig. 3, central). Note that the specimens mentioned here cannot be taken as ‘average’ because the shapes are so disparate in each group. Their relative positions in the stratigraphy are given in Text-figure 2. This demonstrates that samples 2 and 4 come from similar horizons in adjacent sections on Yurtus Mountain. Samples 1 and 3 derive from the same section, but the former is from a stratigraphically lower horizon. Results Frequency histograms of first harmonic amplitudes for each of the four groups are shown in Text- figure 4. Means and standard deviations are shown in Table 1. The data clearly fall into two major 184 PALAEONTOLOGY, VOLUME 40 amplitude of first harmonic amplitude of first harmonic text-fig. 4. Histogram of first harmonic amplitudes, derived from the elliptic Fourier coefficients, for each of the four a priori groups of sclerites. Relationships between these four groups were studied further using an ANOVA (see text). Group means and standard deviations are presented in Table 1. table 1. Means and standard deviations for first harmonic amplitudes derived from the elliptic Fourier coefficients for each of the four a priori groups of sclerites. Relationships between these four groups were studied further using an ANOVA (see text). 1 2 3 4 Mean 123-47 176-32 121-89 179-73 Standard deviation 16-20 28-81 14-01 22-47 groups, the amplitudes of groups 2 and 4 being substantially larger than those of 1 and 3. An ANOVA demonstrated that some of the group means are indeed significantly different from others (degrees of freedom = 3, 486; /-statistic = 253-9 ; probability = 0-000). This result was explored further using a post-hoc Tukey’s test. This confirmed that the amplitudes of groups 1 and 3 are not significantly different from each other, those of groups 2 and 4 are also not significantly different from each other, and that all other comparisons are significant at the > 99-9 per cent, level of confidence. Note that group variances are strongly heteroscedastic, a condition which contravenes one assumption of the ANOVA. Hence data were transformed prior to analysis using Taylor’s power law model (Green 1979). Shape data were examined initially using PCA (Text-fig. 5). These data are displayed as a stereo- plot of the first three principal components designed for viewing with a pocket stereoscope. The x- and v-axes (first and second principal components) may be read without recourse to stereoscopy although interpretation is enhanced by three-dimensional viewing. This plot reveals that the four groups are clustered in morphospace and that there is a greater or lesser degree of overlap between CONWAY MORRIS ET AL. \ LOWER CAMBRIAN CAMBROCLAVES 185 text-fig. 5. Principal components analysis of elliptic Fourier coefficients (harmonics two to six) derived from sclerite outlines. Stereoplots of the first three principal components, which account for 69 per cent, of the total variance. For clarity the data are displayed on two plots. Note that groups are clustered in morphospace, and that all overlap to a greater or lesser extent. all groups. These results were studied in more detail using CVA (Text-fig. 6). This analysis demonstrates very clearly that groups 1 and 3 are indistinguishable on the basis of shape, whereas mean shapes of groups 2 and 4 are significantly different from each other and also from groups 1 and 3. A number of conclusions can be drawn from these results. First, in terms of groups 1 and 3 overlap is almost complete so that they are indistinguishable on the basis of size and shape. This 186 PALAEONTOLOGY, VOLUME 40 first canonical axis text-fig. 6. Canonical variates analysis of elliptic Fourier coefficients (harmonics two to six) derived from sclerite outlines. Plot of first two canonical axes, which account for 97 per cent, of the within- to between- groups variance. Note that groups I and 3 are indistinguishable on the basis of shape, whereas the mean shapes of groups 2 and 4 are clearly distinguished from each other and from groups 1 and 3. The numerals 1-4 represent the position of the group means of TF4. Circles with solid outlines indicate the 90 per cent, tolerance regions for group populations, and those with dashed outlines the 95 per cent, confidence regions on positions of group means. indicates that in Section 5 (Sugaitbulak), from which these groups were collected, there is morphological stasis through the upper part of the Yurtus Formation (Text-fig. 2). Second, groups 2 and 4 are clearly discriminated from each other on the basis of shape. This is largely due to the predominance of sclerites in group 4 with prominent posterior prongs (PL 1, figs 1-3). Moreover, groups 2 and 4 are also discriminated from groups 1 and 3 on the basis of size and shape. In part, this is probably because in the latter two groups the posterior region tends to be reduced in width. The separation in morphospace of these samples, most notably group 4, raises the question of whether all the sclerites should be accommodated in Cambroclavus biconus or should be split into two or more separate taxa. For the moment the former option is preferred, for the following reasons. The distinctiveness of the sclerites in group 4 is due largely to the prong-like extensions. Although very abundant, not every sclerite from this sample possesses this feature, which also occurs sporadically elsewhere in the Yurtus Formation. Moreover, in the cambroclave assemblages from South Australia (Bengtson et al. 1990) the presence of prongs is scattered amongst sclerites from different samples. It is clear that the morphometric methods used here have considerable discriminatory power and that a more extensive survey of cambroclave morphospace would be highly desirable. Several areas stand out as potentially interesting in this regard. For example, the nature of interlocking between adjacent sclerites presumably places constraints on form. Thus in the sclerites of Cambroclavus their dumb-bell shape and presumed articulatory facets are evidently largely determined by the requirements for congruent margins. Further morphometric analysis could be used to generate artificial shapes corresponding to points not filled by real data. Such artificial shapes can help in the CONWAY MORRIS ET AL.\ LOWER CAMBRIAN CAMBROCLAVES 187 visualization and understanding of morphological variation, including the creation of extreme morphologies that exaggerate axes of variation defined by the real data. Our existing data would allow such an exercise in ‘reverse engineering’, although it would be necessary to retain the first harmonic, which in the study reported above was used as a proxy for size. Such a PCA analysis does not give such good shape discrimination between the groups, does not relate to the existing axes of variation, and might be of questionable relevance when so little is yet known about cambroclave palaeobiology. Indeed, even existing palaeontological data may question the utility of such an approach. In the only known examples of relatively extensive sheets of articulated material, which are from Kazakhstan (Mambetov and Repina 1979, pi. 14, figures 6, 8-9), the sclerites appear to have a rather uniform shape. In material from China, however, the most extensive array (Yue 1991, fig. 1.5) consists of two central rows flanked by a row of larger sclerites, but each set appears to show little morphological variability. In assemblages of disarticulated sclerites wide morphological variability seems to be the norm. Assuming that such isolated sclerites derive from a single species and that each individual organism had a variety of morphs, then in principle a shift in sclerite shape across the body could be managed in short distances by minor modifications at each junction. CONCLUSIONS Further understanding of the palaeobiology of the cambroclaves will be hindered until one or more specimens of the entire scleritome are found in a Lower Cambrian Konservat-Lagerstatten, such as the Chengjiang or Sirius Passet biotas. Nevertheless, in the meantime an extensive survey of sclerite morphometries would probably help to identify more natural groupings. The functional significance of the different shapes of sclerite are largely speculative. At present it is envisaged that the sclerites of zhijinitids ( Parazhijinites , Zhijinites) were separately embedded in a tegument, in contrast to the other cambroclaves which formed closely articulated scleritomes (Bengtson el al. 1990; Conway Morris and Chen 1991). A protective role seems to be the most probable function and this presumably would be accentuated by the anterior spines. Nevertheless, the ecological context in which the very slender spines of P. guizhouensis were employed in contrast to the shorter spines of other cambroclaves is speculative. Acknowledgements. Research was supported by the Natural Environment Research Council (GR3/8249), the Royal Society and St John’s College, Cambridge. We thank the following for invaluable assistance: Sandra Last (typing), Hilary Alberti and Sharon Capon (drafting), Ian Marshall (SEM), and Dudley Simons (photography). An anonymous referee provided helpful comments. This is Cambridge Earth Sciences Publication 4670. REFERENCES balsamo, M. and fregni, e. 1995. Gastrotrichs from interstitial fresh water, with a description of four new species. Hydrobiologia, 302, 163-175. bengtson, s., conway morris, s., cooper, b., jell, p. a. and runnegar, b. 1990. Early Cambrian fossils from South Australia. Memoirs of the Association of Australasian Palaeontologists , 9, 1-364. conway morris, s. 1993. The fossil record and the early evolution of the Metazoa. Nature , 361, 219-225. — and chen menge 1991. Cambroclaves and paracarinachitids, early skeletal problematica from the Lower Cambrian of South China. Palaeontology , 34, 357-397. — and peel, j. s. 1995. Articulated halkieriids from the Lower Cambrian of North Greenland and their role in early protostome evolution. Philosophical Transactions of the Royal Society of London , Series B , 347, 305-358. — and chapman, a. j. 1996. Lower Cambrian coeloscleritophorans (Ninella, Siphogonuchites) from Xinjiang and Shaanxi, China. Geological Magazine , 133, 33-51. in press. Lower Cambrian halkieriids and other coeloscleritophorans from Aksu-Wushi, Xinjiang, China. Journal of Paleontology. 188 PALAEONTOLOGY, VOLUME 40 crampton, J. s. 1995. Elliptic Fourier shape analysis of fossil bivalves: some practical considerations. Lethaia, 28, 179-186. crampton, j. s. in press. Biometric analysis, systematics and evolution of Albian Actinoceramus (Cretaceous Bivalvia, Inoceramidae). Monograph of the Institute of Geological and Nuclear Science. ding lianfang, qin hongbin and li yong 1991 [date of imprint 1990], Study on the boundary between Sinian and Cambrian in the south of Shaanxi Province, China. Professional Papers of Stratigraphy and Paleontology , Chinese Academy of Geological Sciences, 23, 96-124. [In Chinese, with English abstract], duan chenghua 1984. [date of imprint 1983]. Small shelly fossils from the Lower Cambrian Xihaoping Formation in the Shennongjia district, Hubei province - hyoliths and fossil skeletons of unknown affinities. Bulletin of the Tianjin Institute of Geology and Mineral Resources , Chinese Academy of Geological Sciences, 7, 143-188. [In Chinese, with English abstract], — and xiao bing 1992. Restudy on the stratotype of Yurtus Formation. Bulletin of the Tianjin Institute of Geology and Mineral Resources, 26-27, 325-352. [In Chinese, with English summary], dzik, j. 1994. Evolution of "small shelly fossil” assemblages. Acta Palaeontologica Polonica, 39, 247-313. elicki, o. 1994. Lower Cambrian carbonates from eastern Germany: palaeontology, stratigraphy and palaeogeography. Neues Jahrbuch fur Geologie und Paldontologie, Abhandlungen, 191, 69-93. — and Schneider, j. 1992. Lower Cambrian (Atdabanian/Botomian) shallow-marine carbonates of the Gorlitz Synclinorium (Saxony/Germany). Facies, 26, 55-66. evans, w. A. 1992. Five new species of marine Gastrotricha from the Atlantic coast of Florida. Bulletin of Marine Sciences, 51, 315-328. ferson, s., rohlf, f. j. and koehn, r. k. 1985. Measuring shape variation of two-dimensional outlines. Systematic Zoology , 34, 59-68. gao zhenjia el al. 1 985c/. The Sinian system on Aksu-Wushi region, Xinjiang, China. Xinjiang People’s Publishing House, Urumqi, 184 pp. [In Chinese, with English abstract], — 1 9856. The Sinian system of Xinjiang {No. 1 stratal summary of Xinjiang). Xinjiang People’s Publishing House, Urumqi, 173 pp. [In Chinese, with English abstract], geyer, g. and elicki, o. 1995. The Lower Cambrian trilobites from the Gorlitz Synclinorium (Germany) - review and new results. Palaontologische Zeitschrift, 69, 87-119. green, r. H. 1979. Sampling design and statistical methods for environmental biologists. Wiley, New York, xi + 257 pp. hou xianguang and Bergstrom, j. 1994. Palaeoscolecid worms may be nematomorphs rather than annelids. Lethaia, 27, 11-17. — and SUN weiguo 1988. Discovery of Chengjiang fauna at Meishucun, Jinning, Yunnan. Acta Palaeontologica Sinica, 27, 1-12. [In Chinese, with English abstract], jiang zhewen 1992. The Lower Cambrian fossil record of China. 311-333. In lipps, j. h. and signor, p. w. (eds). Origin and early evolution of the Metazoa. Plenum, New York, xiii + 570 pp. kuhl, f. p. and giardina, c. R. 1982. Elliptic Fourier features of a closed contour. Computer Graphics and Image Processing, 18, 236-258. landing, e. 1991. Upper Precambrian through Lower Cambrian of Cape Breton Island: faunas, paleoenvironments, and stratigraphic revision. Journal of Paleontology, 65, 570-595. mambetov, a. m. and repina, l. n. 1979. [The Lower Cambrian of Talasskij Ala-Too and its correlation with the sections of Malyi Karatau and the Siberian platform.] 93-138. In Zhuravleva, i. t. and meshkova, n. p. (eds). [ Biostratigraphv and palaeontology of the Lower Cambrian of Siberia .] Nauka, Novosibirsk, 161 pp. [In Russian], nitecki, m. h. 1986. Receptaculitids and their relationship to other problematic fossils. 27-34. In hoffman, a. and nitecki, m. h. (eds.) Problematic fossil taxa. Oxford University Press, New York, xi + 267 pp. qian jianxin and xiao bing 1984. An early Cambrian small shelly fauna from Aksu-Wushi region, Xinjiang. Professional Papers in Stratigraphy and Palaeontology, Chinese Academy of Geological Sciences, 13, 65-90. [In Chinese, with English abstract]. qian yi 1978. The early Cambrian hyolithids in central and south-west China and their stratigraphical significance. Memoirs of the Nanjing Institute of Geology and Palaeontology , 11, 1-38. [In Chinese, with English abstract], — 1989. Early Cambrian small shelly fossils of China with special reference to the Precambrian-Cambrian boundary. Stratigraphy and Palaeontology of Systematic Boundaries in China , Precambrian-Cambrian Boundary, 2, 1-341. — and yin gongzheng 1984. Zhijinitidae and its stratigraphical significance. Acta Palaeontologica Sinica, 23, 215-223. [In Chinese, with English abstract]. CONWAY MORRIS ET AL.\ LOWER CAMBRIAN CAMBROCLAVES 189 — and zhang shiben 1983. Small shelly fossils from the Xihanoping Member of the Tongying Formation in Fangxian County of Hubei Province and their stratigraphical significance. Acta Palaeontologica Sinica , 22, 82-94. [In Chinese, with English abstract]. ROHLF, F. J. 1993. NTSYS-pc ; numerical taxonomy and multivariate analysis system; version 1.80. Exeter Software, Setauket. — and archie, J. w. 1984. A comparison of Fourier methods for the description of wing shape in mosquitoes (Diptera: Culicidae). Systematic Zoology, 3, 302-317. — and bookstein, f. l. (eds). 1990. Proceedings of the Michigan Morphometries Workshop, 1st ed. (Special publication, 2). The University of Michigan Museum of Zoology, Ann Arbor, Michigan, viii + 380 pp. wang wuyan, xiao bing, zhang sengui and zhu zhaoling 1985. Division and correlation of Cambrian System in Aksu-Wushi district of Xinjiang. Xinjiang Geology, 3, 59-74. [In Chinese, with English abstract]. wang yangeng et al. 1984. [The Sinian-Cambrian boundary in central, western and northern parts of Guizhou.] Bulletin of the Institute of Geology, Chinese Academy of Geological Sciences, 10 (Special Issue on the Sinian-Cambrian boundary in China), 91-110. [In Chinese]. white, r. j., prentice, h. c. and VERWIJST, T. 1988. Automated image acquisition and morphometric description. Canadian Journal of Botany, 66, 450-459. xiao bing and duan chenghua 1992. Review of small shelly fauna of Yultus, early Cambrian of Xinjiang. Xinjiang Geology, 10, 212-232. [In Chinese, with English abstract]. yue zhao 1991. Discovery of fused sclerites of early Cambrian Phyllochiton and its relation with zhijinitids. Kexue Tongbao, 1991, 47-50. — and gao linzhi 1992. Paleontology, biostratigraphy and geological significance of the early Cambrian protoconodonts and other skeletal microfossils from Aksu-Wushi region, Xinjiang, China. Bulletin of the Institute of Geology, Chinese Academy of Geological Sciences , 23, 133-160. [In Chinese, with English abstract]. — 1994 [date of imprint 1993], A new early Cambrian species of Tannuolina from Xinjiang region, China. Professional Papers of Stratigraphy and Palaeontology, Chinese Academy of Sciences, 24, 66-78. [In Chinese, with English abstract]. S. CONWAY MORRIS A. J. CHAPMAN Department of Earth Sciences University of Cambridge Downing Street Cambridge CB2 3EQ, UK J. S. CRAMPTON Institute of Geological and Nuclear Sciences PO Box 30368 Lower Hutt, New Zealand XIAO BING Institute of Geology and Mineral Resources Typescript received 18 August 1995 Urumqi, Xinjiang Revised typescript received 13 May 1996 People's Republic of China THE MORPHOLOGY AND SHELL MICROSTRUCTURE OL THE TH ECI DEI DINE BRACH IOPOD ANCORELLINA AGERI FROM THE LOWER JURASSIC OF ARGENTINA by PETER G. BAKER and MIGUEL O. MANCENIDO Abstract. Serial sectioning of complete shells of Ancorellina ageri enables the first description of dorsal valve internal morphology and shell microstructure. The diagnostic description by Mancenido and Damborenea (1990) is inadequate, as the supposedly diagnostic features of the Ancorellina hemispondylium are known to occur in the Aalenian thecidellinid Moorellina. Elowever, the Ancorellina brachidium, consisting of a laterally supported bifurcating column, is currently unique in the Thecideidina. The early ontogeny conforms with the thecideid pattern. The adult brachidium is believed to have supported a ptycholophe and may be interpreted as a precursor of the ramulate condition of lacazellins. Accordingly, the genus is placed in a new subfamily, Ancorellinmae, in the family Thecideidae. The brachidium-supporting pillars resemble the hamate skeletal structures of thecospirellids, introducing the possibility that thecospirellids are close to thecideid ancestral stock. The shell microstructure indicates a phylogenetic link with the Carnian Thecospira haidingeri and probable dispersal from Tethyan faunas early in the Early Jurassic. The first Lower Jurassic thecideidine brachiopod from South America was reported by Mancenido and Damborenea (1990), and was known only from complete shells and separated ventral valves occurring as cryptically cementing epifauna on anthozoan and oyster substrates. Features of the interior of the dorsal valve remained unknown, although, viewed externally, traces of structure visible through the shell suggested a monoseptate condition. The authors did not present a formal systematic diagnosis of Ancorellina because of the essentially palaeobiological/palaeogeographical content of their contribution. Subsequently, serial sectioning has revealed the presence of an unusual brachial skeleton in the genus and a formal diagnosis is now presented. MATERIAL AND METHODS Specimens of Ancorellina ageri were detached from their substrate (underside of scleractinian corals) and prepared using the techniques documented by Baker and Elston (1984, p. 777). Registration of material. The syntypes (MLP12197, 18289-18290, 24470) together with original and duplicate acetate peels of the serially sectioned hypotypes (PB3257-3258) figured in this paper are housed in La Plata Natural Sciences Museum, Argentina. IPalaeontology, Vol. 40, Part 1, 1997, pp. 191-200, 2 pls| © The Palaeontological Association 192 PALAEONTOLOGY, VOLUME 40 SYSTEMATIC PALAEONTOLOGY Order spiriferida Waagen, 1883 Suborder thecideidina Elliott, 1958 Superfamily thecideoidea Gray, 1840 Family thecideidae Gray, 1840 Subfamily ancorellininae subfam. nov. Diagnosis. Forms with median septum resorbed posteriorly; brachidium consisting of an anterior, distally bifurcated median column supported by a pair of anterolaterally placed pillars and directed posteroventrally; lophophore probably ptycholophous; fibrous secondary shell continuous in both valves. Age. Early Jurassic. Genus ancorellina Mancenido and Damborenea, 1990 Derivation of name. From the Latin ancora , after its peculiar anchor-shaped structure. Type species. Ancorellina ageri Mancenido and Damborenea, 1990. Age. Early Jurassic, Late Pliensbachian. Diagnosis. Ancorellinin with relatively large cicatrix and well-developed free ventral wall; ventral interarea reduced; pseudodeltidium indistinct; hinge line short; ventral valve with sessile hemispondylium raised anteriorly to form spoon-like termination; dorsal valve with relatively massive cardinal process and median septum reduced to anteroposteriorly flattened, distally bifurcated column, each limb united laterally with pillar arising from anterior of each brachial cavity; endopunctate. Ancorellina ageri Mancenido and Damborenea, 1990 Plates 1-2; Text-figures 1-3 1990 Ancorellina ageri Mancenido and Damborenea, p. 90, fig. 1. EXPLANATION OF PLATE 1 Figs 1-11. Ancorellina ageri Mancenido and Damborenea, 1990. Upper Pliensbachian; Neuquen, Argentina. 1-4, PB3257; dorsal, lateral, anterior and posterior views, photographic record of hypotype, sectioned adult shell; the apparent ventral umbo is shown by sectioning to be an adherent matrix artefact; x20. 5-8, PB3258; dorsal, lateral, anterior and posterior views, photographic record of hypotype, sectioned presumed juvenile shell; x44. 9, PB3258/8; near transverse section through the ventral valve free ventral wall; showing the granular primary layer (extreme left) and fibrous secondary layer with the early development of tubercles (inner boundary) (section orientation as in Text-fig. 2); x 240. 10, PB3257/42; dorsal valve; showing the ‘chaotic’ disturbance of the fibrous secondary shell layer resulting from close spacing of tubercles (section orientation as in Text-fig. 1 ; section location, right anterolateral sector); x 240. 11, PB3257/32; near transverse section through the dorsal median column showing the oblique secondary fibre orientation; x225. All scanning electron micrographs of gold-coated material; figures 9-11 are of cellulose acetate peels of sectioned specimens. PLATE 1 BAKER and MANCENIDO, Ancorellina 194 PALAEONTOLOGY, VOLUME 40 Type specimens. Holotype not designated; species erected on syntypes MLP12197, MLP18289-18290, MLP24470; hypotypes PB3257-3258. Emended diagnosis. Ancorellina up to about 2-5 mm long, 3 0 mm wide and 1-5 mm thick; sub-circular in outline, commonly with incipient anterior sulcus; pseudodeltidium barely demarcated; hinge line short, about half maximum width of shell; dorsal umbo quite prominent but with no trace of dorsal interarea, valve rather flattened away from umbonal region; endopunctae with restricted distribution. Description. A small thecideid, sub-circular in outline, with a relatively large attachment scar and well- developed free ventral wall, commonly with incipient sulcus, giving a ventribiconvex lateral profile. The ventral interarea is small with an indistinct pseudodeltidium. The overall weak convexity of the dorsal valve is masked by the prominent development of the umbonal region. Endopunctation possibly confined to the dorsal valve. Distribution. Currently known only from the Piedra Pintada area, Neuquen, Argentina. The material was collected from strata of Late Pliensbachian age ( Fanninoceras ammonite Zone, Radulonectites sosneadoensis bivalve Assemblage Zone (Riccardi et al. 1989)). MORPHOLOGY, GROWTH AND SHELL MICROSTRUCTURE Valve characters Ventral valve (Text-figs Ia-y, 2a-g). The pseudodeltidium is not demarcated externally but its presence is marked by a delthyrial notch flanked by well-developed, boss-like, cyrtomatodont teeth. Although the anterior is raised (PI. 2, fig. 2), the posterior of the sessile hemispondylium is sunk into the floor of the valve between thickened dental ridges. Anterior to the hemispondylium, a median ridge extends as far as the base of the free ventral wall, dividing the body cavity overlying the area of attachment into two large shallow oval depressions. The valve peripheral margin is ornamented by tubercles. Dorsal valve (Text-figs 1l-f', 2f-j, 3a-d). The cardinal process is relatively large and incipiently trilobed, and the inner socket ridges are strongly developed although the bridge abutments are small. The subperipheral rim is narrow and ornamented by one or two rows of tubercles. The posterior of the median septum is resorbed and the anterior is extended posteroventrally as an anteroposteriorly flattened column about 05 mm wide and 04 mm high, bifurcated for about half its length. The brachial cavities are devoid of brachial lobes but, from the anterior of each, a pillar about 02 mm thick rises to unite with the distal bifurcation of the median column. The shell is penetrated by irregularly distributed endopunctae (PI. 2, fig. 1). Ontogeny. The ontogeny of A. ageri has been deduced partly from serial sections of a juvenile shell and partly from the preservation of ontogenetic relics buried in adult shell fabric. Although the scarcity of material makes it difficult to be certain, the free ventral wall just beginning to grow away from the substrate, associated with the cicatrix occupying most of the ventral valve surface, and the relatively prominent dorsal umbo of specimen PB3528 (PI. 1, fig. 7), are characteristic of other thecideidines known to be juvenile specimens (Nekvasilova 1967; Baker 1989). The shell microstructure (PI. 1, fig. 9) is essentially the same as that of the adult specimen (PB3527). Therefore, on the basis of morphology and shell microstructure, specimen PB3528 is assumed to be a juvenile A. ageri. Horizontal sections show that in the ventral valve (Text-fig. 2d-e), the robust dental ridges are an early development, but the characteristic hemispondylium and median ridge appear later. Horizontal sections through the dorsal valve (Text-fig. 2f-j), reveal the presence of a short dorsal median septum and no evidence of skeletal brachial supports. Buried ontogenetic relics in the adult shell fabric (PI. 2, fig. 1 ; Text-fig. Ib'-f') suggest that the anterior of the median septum broadened and developed an early sinus typical of the early ontogeny of thecideid genera (Pajaud and Smirnova 1971; Baker and Laurie 1978; Baker and Elston 1984). BAKER AND MANCENIDO, JURASSIC BRACHIOPOD 195 text-fig. 1. Ancorellina ageri Mancenido and Damborenea, 1960. a-f', ‘Wild’ stereomicroscope traces of cellulose acetate peels (out of series 1-50) of serial sections through specimen PB3257. Plane of section horizontal, approximately parallel with the commissural plane. Gross shell fabric mosaic included in dorsal valve traces b'-f', clearly indicating the ontogenetic relic of a broad median septum with sinus. Abbreviations : b.a., bridge abutment; h., hemispondylium ; l.p., lateral pillar; m.c., median column; m.s.r., juvenile median septum relic; s.r. juvenile median sinus relic; t., hinge tooth. Numbers indicate peel cumulative distance in mm from base of free ventral wall (as seen in PI. 1, fig. 3). Scale bar represents 1 mm. Microstructure. In common with other Lower Jurassic thecideidines, fibrous secondary shell forms a continuous lining in both valves. Because of exfoliation loss, the granular primary shell layer (PI. 2, fig. 4) is of indeterminate thickness in the specimens studied but the preservation of growth line 196 PALAEONTOLOGY, VOLUME 40 traces (PI, 1, fig. 3) suggests that it must have been thin, possibly c. 25 pm thick. In the well- preserved fibrous secondary shell, fibres are deflected around closely spaced (rarely > 20 //m apart) fibrous tubercle cores 30-40 //m in diameter. The close spacing of the tubercles, and consequent distortion of the fibrous secondary shell lining associated with them, produces a characteristically ‘chaotic’ gross fabric (PI. 1, fig. 10). In the ventral valve, the tubercle cores are orientated almost parallel with the shell surface and, in anterolateral sectors, the disturbed mosaic associated with them forms an inner layer separated from the primary shell by an outer layer of tubercle-free, more uniformly arranged fibres with a length orientation almost at right angles to those of the inner layer (PI. 2, fig. 3). In the dorsal valve, the tubercle cores are almost perpendicular to the shell surface and the disruption of the fibrous secondary shell affects the whole shell layer (PI. 1, fig. 10). The brachial skeleton is also composed of fibrous shell (PI. 1 , fig. 11) with the fibres apparently obliquely spiralled relative to the median column and pillar axes. Endopunctae have only been recognized in lateral and posterolateral sectors of the dorsal valve. The ventral valve is apparently impunctate. DISCUSSION Morphology In AncorelUna , the form of the hemispondylium with its raised anterior termination (anchor-like structure of Mancenido and Damborenea 1990) is unusual, and the posterior depression into the floor of the ventral valve suggests that the diductor muscle scars were impressed as in Pachymoorellina (Baker 1989). The resorbed median septum succeeded by the ‘three-pronged’ brachidium is reminiscent of an attempt at a perforate Lacazella- like ramulate skeletal support so that, although no trace of lophophore grooves was detected, the configuration of the brachial skeleton suggests that the lophophore was in the form of a ptycholophe. The weakly developed crura were possibly not united to form a bridge. Microstructure The fibrous secondary shell gross mosaic is similar to that described in the Aalenian thecidellinid Moorellina Elliott, 1953 (Baker 1970). Although the fibrous tubercle cores and ‘laminate layering’ typical of the Moorellina free ventral wall (Baker 1970, p. 88) are seen in Ancorellina , in the latter genus the tubercles are not reniform in transverse section. Similarly in the dorsal valve, although the disturbed fibrous secondary mosaic is very similar in both genera, Moorellina has granular calcite tubercle cores. Although the diameter of the tubercles (30^-0 pm) is approximately the same, they are much more closely spaced in Ancorellina (15-20 /an; Moorellina 20-60 pm). In Ancorellina , the close spacing and consequent ‘chaotic’ disruption of the secondary mosaic gives a gross fabric almost identical with that of Tliecospira haidingeri (Suess, 1854) (compare PI. 1, fig. 10 with Benigni and Ferliga 1989, p. 531, fig. 13). EXPLANATION OF PLATE 2 Figs 1-4. Ancorellina ageri Mancenido and Damborenea, 1960. 1, PB3257/46; dorsal valve, showing endopunctae and buried ontogenetic relic of juvenile median septum with sinus; the floor of the sinus has been breached through exfoliation loss of outer shell (section orientation as in Text-fig. 1 ; section location, mid-line approximately 0-8 mm from dorsal umbo); x 240. 2-4, PB3257/20; ventral valve. 2, oblique section through the anterior of the hemispondylium, showing the raised spoon-like termination; the apparent perforation is a function of the orientation of the section (as in fig. 1); x240. 3, showing the differing orientation of secondary fibres in the inner and outer regions of the fibrous secondary shell layer of the ventral valve (section location, free ventral wall, anterolateral sector); x 375. 4, showing junction between the granular primary and fibrous secondary shell layers (section location as fig. 3); x 1500. All scanning electron micrographs of gold-coated cellulose acetate peels of sectioned specimens. PLATE 2 BAKER and MANCENIDO, Ancorellina 198 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Ancorellina ageri Mancenido and Damborenea, 1960. a-j, 'Wild’ stereomicroscope traces of cellulose acetate peels (6-15 out of series 1-15) of serial sections through juvenile specimen PB3258. Plane of section horizontal, relative to the attachment cicatrix; intersecting the commissural plane with an angle of about 5° ventral deflection. Abbreviations: d.r., dental ridge; s., medium septum. Ventral valve stippled. Peel interval approximately 20 pm. Scale bar represents 0-5 mm. text-fig. 3. Ancorellina ageri Mancenido and Damborenea, 1960. a-b, reconstruction of the brachial skeleton of serially sectioned specimen PB3257 based on a vertical plot of the data from acetate peels 20-28 and 30-34 (Text-fig. Ij-w). A, anterior view, and b, lateral view showing the posteroventrally directed distally bifurcated median column and lateral supporting pillars; vertical scale x 3-5 horizontal, c, skewed ‘idealized’ compilation of data from figs a and b showing the brachial skeleton of Ancorellina in oblique profile. Vertical scale x 3 horizontal. D, reconstruction of the interior of the dorsal valve of A. ageri based on superimposition of data from acetate peels 20-50 inclusive. Scale bars represent 0-5 mm. Geographical distribution Although morphological similarity may be explained by the homoeomorphy likely to be encountered in reef-associated faunas (Baker 1984), the physiological processes involved in the secretion of such a closely similar secondary shell fabric in Ancorellina and Thecospira must indicate a close phylogenetic relationship. The occurrence of Ancorellina in the Pliensbachian of Argentina therefore has interesting palaeogeographical implications. Despite their wide geographical separation, it seems that the Argentinian Pliensbachian thecideids are genetically linked with Tethyan thecospirids of Carnian age, although whether the dispersal of the Tethyan populations was via the Hispanic Corridor (Smith and Tipper 1986) (Central Atlantic Seaway of Mancenido and Damborenea 1990) rather than the circum- Antarctic Australasian migration route around southern Gondwanaland, involving ‘island hopping’ (Stanley and McRoberts 1993) or teleplanic larval BAKER AND MANCENIDO, JURASSIC BRACHIOPOD 199 (Newton 1988) strategies, remains unclear. However, as a cemented representative of cryptic epifauna, the opportunities for large jump larval dispersal must have been limited and, as the central Pacific Ocean was at its widest during the Pliensbachian, the Hispanic Corridor route is probably to be preferred. CONCLUSIONS Although the general shell microstructure shows a thecospirid affinity and the anterolaterally placed pillars possibly equate with the hamate structures (Dagis 1973) of thecospirellids, the evidence of the juvenile median septum with an anterior sinus, succeeded by a crudely ramulate brachidium, suggests that Ancorellina is an early thecideid close to the ancestral stock of the Lacazella line. It has been generally accepted (Baker 1990) that the hungarithecids are ancestral to the thecideoids and the current study confirms the relationship with thecospiroids. However, the evidence provided by Ancorellina , viewed in conjunction with the differing thecideid and thecidellinid early ontogenies (Smirnova 1984), raises the possibility that the two thecideoid stocks diverged early and although the thecidellinids remain close to the hungarithecids, the thecideids may be much closer to the thecospirellids than was formerly thought. If this turns out to be the case, the presence of two clearly differentiated thecideoid lines in the Upper Triassic suggests that the separation of thecospiroids and thecideoids occurred much earlier than is currently believed and that the ancestor of the thecideidines should possibly be sought among reef-associated faunas of Permian age. REFERENCES baker, p. G. 1970. The growth and shell microstructure of the thecideacean brachiopod Moorellina granulosa (Moore) from the Middle Jurassic of England. Palaeontology, 13, 76-99. — 1984. New evidence of a spiriferide ancestor for the Thecideidina (Brachiopoda). Palaeontology , 27, 857-866. 1989. Evaluation of a thecideidine brachiopod from the Middle Jurassic of the Cotswolds, England. Palaeontology , 32, 55-68. — 1990. The classification, origin and phylogeny of thecideidine brachiopods. Palaeontology , 33, 175-191. — and elston, d. g. 1984. A new polyseptate thecideacean brachiopod from the Middle Jurassic of the Cotswolds, England. Palaeontology, 27, 777-791 . — and laurie, k. 1978. Revision of the Aptian thecideidine brachiopods of the Faringdon Sponge Gravels. Palaeontology, 21, 555-570. benigni, c. and ferliga, c. 1989. Carnian Thecospiridae (Brachiopoda) from San Cassiano Formation (Cortina D’Ampezzo, Italy). Revista Italiana di Paleontologia e Stratigrafia, 94, 515-560. dagys, a. s. 1973. Ultrastructure of thecospirid shells and their position in brachiopod systematics. Paleontological Journal , 6, 359-369. elliott, g. f. 1953. The classification of the thecidean brachiopods. Annals and Magazine of Natural History, (12), 6, 693-701. 1958. Classification of thecidean brachiopods. Journal of Paleontology , 32, 373. gray, j. e. 1840. Brachiopoda. 151. In Synopsis of the contents of the British Museum, (42nd edition). G. Woodfall, London, 370 pp. mancenido, M. o. and damborenea, s. E. 1990. Corallophilous micromorphic brachiopods from the Lower Jurassic of west central Argentina. 89-96. In MacKinnon, d. l, lee, d. e. and Campbell, j. d. (eds). Brachiopods through time. Balkema, Rotterdam, 447 pp. nekvasilova, o. 1967. Thecidiopsis ( Thecidiopsis ) bohemica imperfecta n. subsp. (Brachiopoda) from the Upper Cretaceous of Bohemia. Sbornik geologikych ved, Paleontologie, 9, 115-136. newton, c. r. 1988. Significance of “Tethyan” Fossils in the American Cordillera. Science, 242, 385-391. pajaud, d. and smirnova, t. n. 1971. Systematics of brachiopods of the suborder Thecideidina. Paleontological Journal, 5, 191-196. riccardi, a. d., damborenea, s. e. and mancenido, m. o. 1989. Lower Jurassic of South America and Antarctic Peninsula. In westermann, g. e. g. and mancenido, m. o. (eds). Jurassic taxa ranges and correlation charts for the Circum Pacific 3. Newsletters on Stratigraphy , 21, 75-104. smirnova, T. N. 1984. [Early Cretaceous brachiopods (morphology, systematics, phylogeny and their significance in biostratigraphy and palaeozoogeography)]. Nauka Press, Moscow, 199 pp. [In Russian]. 200 PALAEONTOLOGY, VOLUME 40 smith, p. L. and tipper, h. w. 1986. Plate tectonics and paleobiogeography : Early Jurassic (Pliensbachian) endemism and diversity. Palaios, 1, 399-412. Stanley, G. d. and mcroberts, c. a. 1993. A coral reef in the Telkwa Range, British Columbia: the earliest Jurassic example. Canadian Journal of Earth Sciences , 30, 819-831. suess, e. 1854. Uber die Brachiopoden der Kossener Schichten. Academie Wissenschaften Wien , Mathematisch Naturwissenschaftliche Klass Denkschrift, 7, 29-65. waagen, w. H. 1882-1885. Salt Range fossils. Productus-limestone fossils (fasc. 2). Brachiopoda. Memoirs of the Geological Survey of India , Palaeontologia Indica , (13), 1, 329-770. PETER G. BAKER Department of Geology University of Derby Kedleston Road Derby DE22 1GB, UK Typescript received 6th December 1995 Revised typescript received 19 February 1996 MIGUEL O. MANCENIDO Department of Invertebrate Palaeontology Museum of Natural Sciences 1900 La Plata, Argentina TWO DEVONIAN MITRATES FROM SOUTH AFRICA by m. ruta and j. n. theron Abstract. The anomalocystitid mitrate Placocystella africana (Gydo and Voorstehoek shales, upper Emsian-lower Eifelian, Bokkeveld Group, Cape Province, South Africa) is redescribcd. The internal anatomy of the head, the tail morphology and part of the nervous system are reconstructed for the first time. P. africana belongs to the family Allanicytidiidae, together with four other Silurian and Devonian species from Gondwana. The monophyly of the Allanicytidiidae is supported by cladistic analysis. A second mitrate from the upper Emsian of the Bokkeveld Group, Bokkeveldia oosthuizeni gen. et sp. nov. from the Gydo Shale, is described. The distinctive ventral head skeleton of this mitrate consists of 22 plates arranged in five transverse rows and is used as a reference to establish plate homologies in the anomalocystitids. The mitrates are calcite-plated deuterostomes of Ordovician to Carboniferous age. In most classifications, these fossils are regarded as echinoderms and are placed in the order Mitrata of the class Stylophora (Ubaghs 1967). All mitrates consist of a massive part, the head or theca, and an articulated appendage, the tail or aulacophore. No signs of radial symmetry are known in mitrates, but whether this is a primitive or derived feature is disputed. The anatomy, orientation and systematic position of mitrates are strongly debated. Three schools of thought interpret these fossils in different ways. These different interpretations are referred to in the literature as the aulacophore theory of Ubaghs (1967 et seq.), the calcichordate theory of Jefferies (1967 et seq.) and the stele theory of Philip (1979, 1981). Reviews of these theories are found in Philip (1979), Chauvel (1981), Ubaghs (1981), Jollie (1982), Kolata and Jollie (1982), Jefferies (1986), Cripps (1991), Kolata et a/. (1991) and especially Peterson (1994, 1995), who has recently criticized the calcichordate theory of Jefferies by testing its predictions in the context of a higher level phylogeny of the deuterostomes. Two groups of mitrates, the Mitrocystitida and the Anomalocystitida (Caster 1952), deserve particular attention, since for Jefferies (1986, 1991), whose calcichordate theory is largely followed here, the mitrocystitid and anomalocystitid mitrates are stem-group craniates (see also Jefferies and Lewis 1978; Craske and Jefferies 1989; Cripps 1990; Beisswenger 1994). The mitrocystitids and the anomalocystitids would share with craniates a lateral line system and dorsal, touch-sensory branches of the trigeminal nerves (Jefferies 1986). The mitrocystitids are likely to represent a paraphyletic group (Craske and Jefferies 1989). The monophyletic Anomalocystitida are characterized mainly by the possession of two articulated spines and by a regular arrangement of the head plates. The anomalocystitids from the Southern Hemisphere are of special interest, since most of them are comparatively less well known than the boreal species and no transitional forms are known that bridge the morphological gap between these two groups (but see Haude 1995). In his extensive revision of the fossil faunas from the Bokkeveld Group of South Africa, Reed (1925) described a poorly preserved mitrate collected near Buffelskraal (Cape Province) and named it Placocystis africanus. This was the first mitrate recorded from the Southern Hemisphere. However, the choice of both the generic and the specific name was unfortunate, as Placocystis is an invalid emendation of Placocystites de Koninck (Bather 1900) and also a junior synonym of Enoploura Wetherby (Haeckel 1896). Furthermore, Placocystis is a feminine noun, whereas the adjective africanus is masculine. In 1936, Rennie described the mitrate Placocystella capensis on the IPalaeontology, Vol. 40, Part 1, 1997, pp. 201-243, 8 pis] © The Palaeontological Association 202 PALAEONTOLOGY, VOLUME 40 Atlantic Ocean 2o°e 22 °e 24 °e 1 1 1 text-fig. 1 . Distribution of the Ceres Subgroup, with localities of collecting sites. FORMATION thick- AGE ness (m) KAROOPOORT 50 BIDOUW OSBERG 55 GIVETIAN o KLIPBOKKOP 170 _l SUBGROUP WUPPERTAL 65 UJ > 3 WABOOMBERG 200 m o BOPLAAS 30 DO o CERES TRA-TRA 85 EIFELIAN o CD HEX RIVER 100 SUBGROUP VOORSTEHOEK 115 GAMKA 135 GYDO 160 EMSIAN text-fig. 2. Stratigraphical column of the Bokkeveld Group. basis of two specimens from Gamka Poort (Cape Province). For Rennie, Placocystella capensis and " Placocystis' africanus were likely to belong to the same genus and perhaps to the same species. Reed’s and Rennie’s species were assigned to the family Placocystidae by Caster (1952), who later (1954) concluded that they were congeneric. In his revision of the Allanicytidiidae Caster and Gill, 1967, Caster (1983) mentioned Placocystella but did not include it in this family. Ubaghs (1967) diagnosed and figured this genus following Caster’s 1954 paper. Derstler (1979) placed Placocystella capensis in the Mitrocystitida ( sensu Caster 1952), presumably because oral spines are not preserved in Rennie’s specimens, and kept it separate from ‘ Placocystis' africanus , which he assigned to the Anomalocystitida (Reed’s specimen shows spines). Placocystella capensis was reported from the two lowermost formations of the Bokkeveld Group by Oosthuizen (1984). He presumably referred to RUTA AND THERON: DEVONIAN MITRATES 203 text-fig. 3. Placocystella africana (Reed, 1925). Reconstruction of the external aspect. A, left lateral view; b, dorsal view; c, right lateral view; D, ventral view. 204 PALAEONTOLOGY, VOLUME 40 the two lowermost shaly units, the Gydo and the Voorstehoek shales, as mitrates are not known from the Gamka Sandstone. Parsley (1991) considered ‘ Placocy stis' africanus and Placocy Stella capensis as separate species and regarded them as possible members of the Allanicytidiidae, despite the incomplete knowledge of their anatomy. The present work confirms that Reed’s and Rennie’s species are the same. In view of the nomenclatural problems outlined above, the denomination for the South African mitrate is Placocy Stella africana (Reed, 1925). Placocystella capensis Rennie, therefore, becomes a subjective junior synonym. In this paper, the reconstructions and descriptions of the mitrates and the cladistic analysis are by M. Ruta. The review of the Bokkeveld Group, the stratigraphical column and the outcrop map are by J. N. Theron. Repositories: SM, Sedgwick Museum, Cambridge; SAM, South African Museum, Cape Town; JJS, Savill collection, Airton, Skipton, North Yorkshire; BMNH, The Natural History Museum, London. The Savill collection was kindly donated by Mr Jeremy J. Savill to the Department of Palaeontology of The Natural History Museum, London. The specimens in this collection are identified with their former catalogue numbers in parenthesis and by The Natural History Museum registration numbers. REVIEW OF THE BOKKEVELD GROUP OF SOUTH AFRICA Introduction. Although numerous fossils from the Bokkeveld Group were described during the first part of the last century (Thom 1830; Bain 1856; Salter 1856), it was only after 1895 that extensive research yielded a more complete picture of the distribution and stratigraphical setting of the Bokkeveld beds. Their Devonian age was confirmed by comparing their fossil faunas with those from South America and Europe (Corstorphine 1896; Clarke 1913; Reed 1925). The Bokkeveld Group forms part of the clastic Cape Supergroup, which extends for 800 km eastwards and 300 km northwards of Cape Town. The Cape Supergroup is subdivided into the basal, mainly arenitic Table Mountain Group; the middle, markedly argillaceous Bokkeveld Group; and the upper, more arenitic Witteberg Group. Stratigraphical setting. The Bokkeveld Group consists of cyclical sequences of mainly argillaceous units alternating with arenaceous horizons. Each of these litho-units has been given formational status. The six lower formations, recognizable throughout the outcrop area, are referred to as the Ceres Subgroup (Text-fig. 1). In the west, the upper part of the sequence, designated as the Bidouw Subgroup, comprises five formations (Text-fig. 2). In the east, the laterally equivalent Traka Subgroup consists of three formations (Theron 1972; Theron and Johnson 1991). The total sequence is much thicker in the east than in the west. Weathering of the sequence resulted in a hogsback topography: the more resistant arenitic units create ridges, whereas the softer, intervening argillaceous units weather mainly recessively and are less well exposed. The sequence represents five major, upward-coarsening, superimposed cycles, gradually replaced southwards by a relatively homogeneous mudstone-siltstone sequence. The decrease in the amount of coarse elastics is linked with a progressive thickening of the argillaceous units towards the south. The arenaceous units vary from fine-grained quartz arenites to horizontally laminated or cross-bedded immature arkosic arenites. The argillaceous units consist of dark grey EXPLANATION OF PLATE 1 Figs 1-4. Placocystella africana (Reed, 1925). Specimens in figs 1 and 2 from localities K, Klipfontein, and B, Lakenvalleidam, respectively; Ceres Subgroup, Voorstehoek Shale; specimens in figs 3 and 4 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale. 1, latex cast of SAM 0135, centro-dorsal plates, upper lip and stereom structure of oral spines. 2, latex cast of SAM 0036, dorsal aspect of V6, partial head frame, dorsal fore-tail plates and anterior styloid blade. 3, latex cast of BMNH EE 5651 (JJS4), partial dorsal head skeleton and plate n. 4, latex cast of SAM 0113, anterior half of dorsal head skeleton and oral spines. All x 5. PLATE 1 RUTA and THERON, Placocystella 206 PALAEONTOLOGY, VOLUME 40 shale, mudstone and siltstone with thin intercalated lenses of fine- to medium-grained lithic sandstone. Little attention was paid in the past to the stratigraphical occurrence of the Bokkeveld fossils, and for many years no serious attempts were made to detect their zonal distributions. However, geographical differences were noted (Schwarz 1906; Theron 1972; Oosthuizen 1984). For example, brachiopods and echinoderms are commoner in the west, whereas conulariids, corals and hyoliths are prevalent in the east. Bryozoans, fish and ostracodes are rare throughout. Marine invertebrates are particularly common in the Ceres Subgroup, but occur up to the Klipbokkop Formation in the west and the Karies Formation in the east. Although present throughout the sequence, plant and trace fossils are commoner in north-western outcrops. Certain taxa occur most commonly in particular lithologies (Theron 1972; Oosthuizen 1984): trilobites and cephalopods occur mainly in argillaceous horizons; brachiopods, gastropods and bivalves can be found both in arenaceous and in argillaceous units. However, some brachiopod genera seem to have preferred sandy sediments and the infaunal bivalves are largely confined to shaly units. Crinoids are found in a wide variety of sedimentary rocks, varying from mudstone to lithic arenites. Stalk fragments are abundant, but well preserved crinoids are mainly confined to fine-grained units. This is true also for ophiuroids and blastoids, which, although rare, occur mainly in mudstone and shales or in silty shales. The overall decrease in invertebrate fossils in a northward direction, where the argillaceous units become sandier, suggests a shallowing of the basin in that direction. A decrease occurs also southwards, towards the deeper part of the basin (Theron 1970, 1972; Theron and Loock 1988). Palaeogeographical setting. The Bokkeveld sedimentary strata record the most dynamic phase of the Cape basin development, when tectonic unrest and accelerated downwarp evolved at the Pragian- Emsian transition. The vertical stacking of the upward-coarsening sequences implies tectonically controlled regressions and transgressions (Theron 1972; Tankard and Barwis 1982). The five major cycles record the progradation of lobate, wave-dominated deltas along a coastline of moderately high marine energy (Tankard and Barwis 1982; Theron and Loock 1988). The nearshore deposits grade southwards into thick shelf mudstones. The greatest thickness is towards the eastern Cape and reflects increased downwarping in that direction. An idealized Bokkeveld sequence consists of sediments of the shelf, delta slope and delta platform environments deposited successively during the constructional phase of delta growth and is overlain by nearshore marine reworked deltaic deposits produced during the destructional phase of delta development (Tankard and Barwis 1982). The marked change from a few thousand metres of supermature sands of the Table Mountain Group to the predominantly muddy sediments of the Bokkeveld Group throughout the Cape basin represents an overall northward advance of the shoreline and a progression of the Gydo shelf and delta slope sediments across the sand-shoal deposits of the Rietvlei Formation in the early Devonian. The Bokkeveld mud and siltstone deposits are dark grey and preserve scattered external and internal moulds of a rich invertebrate fauna. Coquinites, where present, are relatively thin, but laterally persistent. As water depth increased, these deposits became rarer. Ebbing storm surge currents collected shells and sediment from the sea floor and carried them seawards. Eventually, hollows and storm-generated channels entrapped them in large quantities (Hiller and Theron 1988). Storm activity is also well documented in the overlying delta platform sediments, associated with the distributary mouth bar, the interdistributary bay and the tidal flat (especially in the northern facies; Theron and Thamm 1990). These sedimentary rocks consist mainly of sandstones, siltstones and mudstones. Fossil invertebrates are occasionally found in thick lenticular coquinites. Plant remains are rare. The progressive upward shallowing reflected by these environments caused impoverishment of the invertebrate fauna, as fossils are rather sparse or absent in the marine reworked sands of the delta platform. Wave and tidal activity created interspersed barrier washover sheets as well as tidal inlet and channel filling sequences. Post-mortem transport of the organisms was generally limited. The mechanical damage to shells is minimal. Disarticulated valves occur commonly, but they rarely display signs of abrasion or RUTA AND THERON: DEVONIAN MITRATES 207 text-fig. 4. Placocystella africana (Reed, 1925). Reconstruction of the external aspect, a, anterior view (oral spines omitted); b, posterior view (tail omitted). breakage. Rapid burial of individuals of various species in life position, exquisitely preserved in the Gydo and Waboomberg formations, indicates the action of gentle currents at the depositional site. A sudden influx of fine sediment, either carried by delta-feeding rivers or produced by storms generating a blanket of wave-stirred mud, may have occurred. Palaeoecology. The Bokkeveld benthic community structures were discussed briefly by Boucot (1971) and Hiller and Theron (1988). A number of communities can be correlated with the depositional sub-environments of the delta complex. The tidal flat community was dominated by inarticulate brachiopods and infaunal bivalves and inhabited the sheltered, back-barrier environment. The distributary mouth bar community consisted mainly of brachiopods and occupied the relatively turbulent shallow water setting at the seaward edge of the delta platform. The delta slope community was more diverse. The brachiopods were still dominant, but infaunal bivalves, gastropods, cricoconarids and crinoids were also present. Trilobites were an important component of this community. The shelf community contained the most diverse fauna of all, with brachiopods (less than half of the total assemblage), abundant trilobites, bivalves and gastropods and a significant proportion of echinoderms, hyoliths, corals, bryozoans, conulariids and cephalopods. In the Gydo and Gamka formations, all the above-mentioned communities are well represented. Shelf and slope communities are present in the Voorstehoek Formation. Shelf communities have also been identified in western outcrops of the Waboomberg Formation. Fossils are generally too rare for palaeoecological studies in many of the other formations. The mitrates described here occur largely in silty mudstones of the Gydo and Voorstehoek formations and belong to the shelf community. This inhabited a deep water environment, off" the front of the growing delta and beyond the effects of waves and tides, which explains the deposition of fine sediment. The taxa are represented by numerous suspension feeders (brachiopods, crinoids, corals, bryozoans, epifaunal bivalves, some infaunal bivalves), a significant number of deposit feeders (some trilobites and ophiuroids, epifaunal hyoliths, most infaunal bivalves and, perhaps, mitrates), possible herbivores (some gastropods) and predators (some cephalopods). SYSTEMATIC PALAEONTOLOGY Superphylum deuterostomia Grobben, 1908 (Stem group of the Craniata?) Family allanicytidiidae Caster and Gill, 1967 Diagnosis. Anomalocystitid mitrates characterized by the presence of four paired marginal ventral plates surrounding a large, centro-ventral element; plates e and 6 not sutured with each other mid- ventrally; upper lip plates highly modified to form a flexible articulation with the centro-dorsal and the antero-lateral dorsal marginal plates; styloid with transversely expanded blades. 208 PALAEONTOLOGY, VOLUME 40 text-fig. 5. Placocystella africana (Reed, 1925). a, dorsal head plate nomenclature; b, ventral head plate nomenclature. Genus placocystella Rennie, 1936 Type species. Placocystella capensis Rennie, 1936 (= Placocystella africana Reed, 1925). Placocystella africana (Reed, 1925) Plates 1-8; Text-figures 3-5, 7-21 1925 Placocystis africanus Reed, p. 30, pi. 4, fig. 1. 1932 Placocystites africanus (Reed); Dehm, p. 65. 1936 Placocystis africanus Reed; Rennie, p. 273, pi. 31, fig. 1. 1936 Placocystella capensis Rennie, p. 269, pi. 31, figs 3-9. 1941 Placocystis africanus Reed; Chauvel, p. 217. 1941 Placocystella capensis Rennie; Chauvel, p. 217. 1952 Placocystis africanus Reed; Caster, p. 19. 1952 Placocystella capensis Rennie; Caster, p. 19. 1954 ‘ Placocystis ’ africanus Reed; Caster, p. 145, fig. 3a-c, 4; pi. 8, figs 5-8. 1954 Placocystella capensis Rennie; Caster, p. 145, fig. 3d-f. 1967 Placocystella capensis Rennie; Ubaghs, p. 561, fig. 357, 5. 1979 ‘ Placocystis africanus' Reed; Derstler, p. 102. 1979 Placocystella capensis Rennie; Derstler, p. 102. 1983 Placocystella sp. Rennie; Caster, p. 328. 1984 Placocystella capensis Rennie, pars; Oosthuizen, p. 139. 1989 Placocystella sp. Rennie; Craske and Jefferies, p. 95. 1990 Placocystella capensis Rennie; Cripps, p. 59. 1991 Placocystis africanus Reed; Parsley, p. 16. 1991 Placocystella capensis Rennie; Parsley, p. 16. 1994 Placocystella capensis Rennie; Cripps and Daley, p. 125. Holotype , type locality and type horizon. SM A. 3044; Buffelskraal, Hex River Pass, Gydo Shale; 33°24'S, 19°48'E (Text-fig. 7a). Diagnosis. Plate D10 as long and half as wide as D12; anterior margin of plates V6 and V9 asymmetrical and visible in dorsal view; oral spines robust and cigar-like; riblets on plates Dl, D2, D7, D8, V10, V12, V14, V15, V19, p, e and 9\ lateral ribs on e, 9, V15 and V19; transverse keels on ^ and 6\ anterior, fan-like styloid blade with radial thickenings; posterior styloid blade parabolic in outline and smooth; first hind-tail ossicle more robust than successive ossicles and with fine striations; successive ossicles with knobs; hind-tail plates with irregular, mainly transverse ridges and a dorsal, horizontal keel. RUTA AND THERON: DEVONIAN MITRATES 209 text-fig. 6. a-b, Bokkeveldia oosthuizeni gen. et sp. nov.; A, drawing of the holotype; b, nomenclature of the ventral head plates, c-f, nomenclature of the dorsal head plates in cornutes and mitrates; c, the cornute Reticulocarpos hanusi ; d, the mitrate Chinianocarpos thorali; e, the mitrate Mitrocystites mitra\ f, the mitrate Milrocystella incipiens (c-F redrawn and simplified after Jefferies 1986). Material , localities and horizons. SAM 0002-0003, 0060-0061, 0102-0104; locality D, Gydopas, Gydo Shale; 33°13'55"S, 19°19'35"E. SAM 0072; locality L, Doornkloof, Gydo Shale; 33°23'20"S, 21°08'45"E. SAM 0009, 0011-0015, 0085-0086, 0106-0110, 0112-0114; locality A, Klipfontein, Voorstehoek Shale; 33°21'20"S, 19°30'45"E. SAM 0049-0051, 0092, 0094-0095, 0101, 0132, 0135; locality K, Klipfontein, Voorstehoek Shale; 33°21'25"S, 19°30'25"E. SAM 0035-0036; locality B, Lakenvalleidam, Voorstehoek Shale; 33°19'55"S, 19°33'1 5"E. SAM 0139; locality J, Lakenvalleidam, Voorstehoek Shale; 33°19'55"S. 19°33'I5"E. BMNH EE 5648-5649 (JJS1-2), BMNH EE 5651 (JJS4), BMNH EE 5653-5654 (JJS6-JJS7), BMNH EE 5656 (JJS9), BMNH EE 5658 (JJS11), BMNH EE 5660 (JJS13). BMNH EE 5662-5667 (JJS15-20); Swaarmoed Pass, 210 PALAEONTOLOGY, VOLUME 40 Ceres, Voorstehoek Shale; 33°2r20''S, 19°30'40"E. BMNH EE 5657 (JJS10); Sewenweekspoort, Voorstehoek Shale; 33°21'50"S, 21°22'E. BMNH EE 5668 (JJS21); Tunnel Siding, Voorstehoek Shale; 33°25'S, 19°46'E. BMNH E23684-23685 (plaster replicas of SAM 9701 and SAM 9700 respectively; Gamka Poort, Gydo Shale; 33°18'S, 21°38'E). Most of the South African Museum specimens were collected by Dr R. P. S. Jefferies, Mr J. J. Savill and Dr J. N. Theron in January 1993. Methods of study and plate terminology. The fossils were cleaned in ethanolamine thioglycollate. The reconstructions are based mainly on camera lucida drawings of latex casts coated with ammonium chloride to accentuate relief. In most mitrocystitids and anomalocystitids, the dorsal head skeleton is morphologically conservative, but the ventral head plates vary in number, arrangement, shape and position. Here, the same notation is given to putatively homologous plates in different species. In the case of Placocystella africana , D followed by a numeral applies to the centro-dorsal and to those marginal plates believed to correspond with like-named plates in Mitrocystites mitra and Mitrocystella incipiens (Dl, D2, D7, D8, D10 and D12; Text-fig. 6e-f), whereas the lower-case h, i, a, b, c and d apply to those marginal plates which are perhaps homologous with their namesakes in Reticulocarpos hanusi (Text-fig. 6c; discussion in Jefferies and Prokop 1972). Most of the ventral head plates are labelled as V followed by a numeral, a new system based on the ventral skeleton of Bokkeveldia oosthuizeni gen. et sp. nov. (V6, V9, V10. V12, V14, V15 and V19; Text-fig. 6a-b). Exceptions are plates p, homologous with their namesake in Barrandeocarpus norvegicus (Craske and Jefferies, 1989), and e and 6 , possibly homologous with like-named elements in Prokopicystis tnergli (Cripps, 1989). DESCRIPTION External anatomy of the head Dorsal skeleton (Text-figs 3b, 5a, 7b-c, 8-9, 10a; PI. 1, figs 1-4; PI. 2, figs 2-4; PI. 4, figs 2-3, 6; PI. 6, figs 5-6). The head, elliptical in plan view and bilaterally symmetrical, is mostly flat dorsally. The dorsal skeleton consists of two centro-dorsal and 1 I marginal plates (Text-fig. 5a). The postero-median half of Dl and D8 and the posterior third of i and h slope gently downward and laterally. All marginals except n form pairs of plates along the lateral margins of the head. Each member of a pair is almost a mirror image of the opposite plate. Plates i and h form most of the postero-dorsal skeleton and meet along a straight, mid-dorsal suture. Their postero-median margins form a ridge, underneath which is the dorsal half of the posterior head excavation (Text-figs 8, 9a; PI. 1, figs 2-3; PI. 2, figs 2-3; PI. 4, figs 3, 6; PI. 6, figs 5-6). The lateral margins of i and h are gently sinuous. The anterior margin of i is divided into a left shorter and a right longer segment sutured with D10 and D12 respectively. The anterior margin of h contacts D12 and is concave rearward medially and straight laterally (Text-figs 3b, 5a, 9a, 10a; PI. 1, figs 1-4; PI. 2, fig. 3; PI. 4, fig. 2; PI. 6, fig. 5). The postero- lateral margins of i and h are rounded and contribute to the postero-lateral angles of the head. The outer margins of plates i, Dl, D2 and a on the left, and of plates h, D8, D7 and d on the right are projected downward and form the lateral head walls. The anterior part of the dorsal skeleton forms a right angle with the lateral walls. At the level of Dl and D8 this angle becomes acute. This slope change begins about half-way along the lateral walls. Dl and D8 are triangular elements. The anterior margin of Dl is concave rearward and medianward. The anterior margin of D8 is concave rearward and medianward anteriorly, and forward and rightward posteriorly EXPLANATION OF PLATE 2 Figs 1^1. Placocystella africana (Reed, 1925). Specimens in figs 1, 3^1 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 2 from locality D, Gydopas; Ceres Subgroup, Gydo Shale. 1, latex cast of SAM 0106, partial ventral steinkern, partial ventral surface of dorsal head skeleton, external aspect of some ventral head plates and position of the anterior boundary of the posterior coelom. 2, latex cast of SAM 0060, rearmost part of the dorsal head skeleton. 3, latex cast of SAM 0110, oral spines and partial dorsal head skeleton; a button-like lump is visible on the partially exposed dorsal surface of VI 5. 4, latex cast of SAM 0108, partial of the dorsal head skeleton, left oral spine and part of the lower lip. All x 5. PLATE 2 RUTA and THERON, Placocystella 212 PALAEONTOLOGY, VOLUME 40 text-fig. 7. Placocystella africana (Reed, 1925). a, holotype, SM A. 3044; Buffelskraal, Hex River Pass; Ceres Subgroup, Voorstehoek Shale; natural external mould of ventral head skeleton and partial natural internal mould of mid- and hind-tail ; x 5. b, BMNH E23684, plaster replica of SAM 9701 ; northern entrance of Gamka Poort; Ceres Subgroup, Gydo Shale; external mould of dorsal head skeleton; x 4. c, BMNH E23685, plaster replica of SAM 9700; horizon and locality as b; external mould of dorsal head skeleton and partial ventral internal mould; x4. RUTA AND THERON: DEVONIAN MITRATES 213 text-fig. 8. Placocystella africana (Reed, 1925). a, latex cast of BMNH E23684; x 4. B, interpretative sketch, c, latex cast of BMNH E23685; x 4. d, interpretative sketch. (Text-figs 3b, 9a, 10a; PI. 1, fig. 1 ; PI. 6, figs 5-6). D2 and D7 are trapezoidal. The median margin of D7 follows the curvature of the right head margin. The straight, median margin of D2 is not parallel to the body axis (Text- fig. 9; PI. 1, figs 2-4). 214 PALAEONTOLOGY, VOLUME 40 The sub-quadrangular plates a and d carry the articulation of the oral spines antero-laterally. The median part of the anterior margin of a forms a short junction with n (PI. 1, figs 1, 4; PI. 4, fig. 6), whereas its lateral part contacts b. The junction between n and d (Text-figs 9a, 10a; PI. 1, figs 3^1) corresponds to the median part of the anterior margin of d (PL 5, fig. 6). The lateral part of this margin contacts c. Plates b, n and c form the upper lip. Plate b is wedge-shaped and its longitudinal axis faces forward and leftward. Plate c seems to be shorter than b and is more equilateral. Its longitudinal axis faces forward and rightward (Text-figs 9a, 10a, 1 1 ; PI. 1, fig. 1 ; PI. 2, fig. 3). Plate n forms most of the upper lip and has a slightly convex dorsal surface. Its left margin contacts b and a, whereas its right margin contacts c and d. Its anterior margin is convex outward. Its posterior margin is divided into two unequal segments forming a triple junction with DIO and D12: the left segment is sinuous, whereas the right segment is straight (Text-fig. 9a; PI. 1, figs 1, 3-4). The junctions of b, n and c with a, DIO, D12 and d are cylindrically rounded (Text-fig. 9a; PI. 1, figs 1, 3) and perhaps allowed the upper lip to be lifted in life. DIO is five- or six-sided; D12 is seven-sided. The D10/D12 suture, divided into a sinuous anterior and a straight posterior part, runs from the mid-point of the posterior margin of n to the anterior margin of i. Ventral skeleton (Text-figs 3d, 5b, 10b; PI. 2, fig. 1 ; PI. 3, figs 1-4; PI. 4, figs 4—5; PI. 5, figs 2-5; PI. 8, fig. 2). The ventral skeleton is a rigid, gently convex structure of ten plates, nine of which are marginals (Text-fig. 5b). All ventral marginals except p form pairs. Each member of a pair is almost a mirror image of the other. VI 2, sometimes found articulated only with plates e, 9 and p, is diamond-shaped and almost bilaterally symmetrical (Text-fig. 10b; PI. 3, figs 2—4; PI. 5, figs 3-5; PI. 8, fig. 2) and contacts the ventral marginals along its nine sides. In adults of P. africana, the posterior margin of V12 is fused with p (PI. 5, figs 4-5), but one small, presumably young individual seems to show a distinct p/V12 suture (PI. 5, fig. 3). Plates s and 6 form the postero-ventral skeleton and part of the posterior head surface. In lateral view, their ventral curvature is more prominent than that of other ventral plates. Plates e and 0 contact: h and i posteriorly; the postero-lateral sides of V12 and the lateral margins of p medially; and VI 5 and V19 anteriorly along two sinuous sutures. V15 and V19, the smallest ventral marginals, are trapezoidal. V10 and V14 are slightly larger than V15 and V19, but are smaller than V6 and V9. V6 and V9 occupy the anterior third of the head floor and contact each other along a straight, mid-ventral suture. Postero-medially, they are sutured with the anterior sides of V12. These are not symmetrical, as the right side is slightly longer than the left. The anterior margins of V6 and V9, visible in dorsal view, show a gentle anterior surface, a rounded upper edge and a steep posterior surface (Text- figs 8-9, 11; PI. 1, figs 2, 4; PI. 4, fig. 3; PI. 5, fig. 6). Ornament (Text-figs 3-4, 7b-c, 8, 10; PI. 3, figs 1—4; PI. 5, figs 2-5; PL 8, fig. 2). This consists of ribs, riblets and broken ribs. The ribs are transversely elongate with an anterior steeper and a posterior gentler slope (cuesta-like profile). The distance between each rib is about the same in individuals of different sizes, which are presumed to be of different ages, suggesting that new ribs were formed as the head grew. The riblets are sub-rectangular to crescentic sculptures, densely grouped and particularly numerous on VI 2. Like ribs, they are cuesta-like. Their anterior margin shows minute denticles. The broken ribs are irregular lines marking the spatial transition from ribs to riblets and are sometimes visible on the middle and posterior parts of the lateral head walls (Text-fig. 10; PL 3, fig. 2). Widely spaced riblets on Dl, D2, D7 and D8 (Text-fig. 10a) sometimes become confluent and give rise to irregular, sinuous ribs. Along the median margins of Dl, D2, D7 and D8 and behind the anterior margins of D2 and D7 the ornament is absent. D10, D12, a, b, n, c, d, h and i are usually smooth, but some specimens show riblets on a and d (Text-fig. 8). V6, V9 and the anteriormost part of V12 are smooth. V10 and V14 have few widely spaced riblets stopping behind their anterior margins. The anterior half of both VI 5 and V19 carries widely spaced riblets, whereas the postero-lateral half shows ribs. These become broken and sinuous medially and continue on the anteriormost surfaces of e and 9. V12 has closely spaced riblets confined to its posterior two-thirds (Text-fig. 10b; PL 3, figs 1^1; PL 5, figs 3-5). Two robust keels run from the postero-lateral angles of p to the centre of the posterior third of e and 9 (PL 5, figs 3-5). The external surface of these two plates in front of and behind the keels is smooth and elliptical in outline and continues medially on plate p. Riblets cover a narrow transverse area just behind the anterior margin of p. The lateral surfaces of e and 9 bear ribs which become wavy and irregular anteriorly. Near the anterior half of the median margins of these two plates the ornament consists mainly of riblets. RUTA AND THERON: DEVONIAN MITRATES 215 text-fig. 9. Placocystella africana (Reed, 1925). A, latex cast of BMNH EE 5649 (JJS2); Swaarmoed Pass; Ceres Subgroup, Voorstehoek Shale; almost complete dorsal head skeleton and antero-lateral part of dorsal surface of V9; x 5. B, latex cast of SAM 0106; locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; partial dorsal head skeleton, anterior margin of ventral head skeleton and oral spines; x 5. Oral spines (Text-figs 3, 5, 7a, 9b, 11; PI. 1, figs 1, 4; PI. 2, figs 3-4; PI. 3, figs 3 — 4 ; PI. 4, figs 1, 3-6; PI. 5, fig. 6; PI. 6, figs 5-6; PI. 8, fig. 2). The spines, elliptical in cross section throughout their length, taper distally. The proximal dorsal surface of each spine slopes gently rearward and downward towards a dorso-ventrally elongate socket. Two facets are present at the base of each spine. One of these facets is lateral and twice as wide as the other, which is median to the socket (Text-fig. 11; PI. 3, fig. 4; PI. 5, fig. 6; PI. 6, fig. 6). The spines are articulated to the thickened antero-lateral angles of a and d (Text-figs 3b, 4a, 9, 11; PI. 4, fig. 3; PI. 5, fig. 6; PI. 6, fig. 6). The attachment areas comprise a flat, sub-circular areola with a stout process which occupies a median position on it. The process carries a low platform, on top of which is a smooth, dorso-ventrally elongate tubercle which fits into the spine socket. The spines probably swept mainly horizontally in life, but the plane of sweep could probably be adjusted vertically to some extent. Tail As in all mitrates, the tail of P. africana consists of three parts: fore-, mid- and hind-tail, in order of increasing distance from the head. 216 PALAEONTOLOGY, VOLUME 40 Fore-tail (Text-figs 3, 10a; PI. 1, fig. 2; PI. 4, fig 3; PI. 6, figs 1-5; PI. 8, fig. 4). The fore-tail skeleton consists of calcitic rings (the highest recorded number is six) enclosing a lumen, presumably occupied in life by muscle. Each ring is made of four rigidly sutured plates: left and right dorsal, and left and right ventral. The ventral and dorsal plates of one side meet along a flat, horizontal suture. Each ventral plate consists of two parts, gently convex outward and smooth-surfaced. The smaller, lower part forms most of the ventral aspect of the fore-tail, whereas the larger, upper part contributes to most of its lateral surface (PI. 6, fig. 1). Like the ventral plates, each dorsal plate shows two parts. The lower is gently convex outward and contributes to the lateral surface of the fore-tail. The upper is slightly concave outward. The median part of the upper surface of each dorsal plate is upturned and forms a mid-dorsal keel with its antimere. The dorsal plates have a rounded, postero-median angle and a low posterior ridge (PI. 6, fig. 3). The fore-tail was flexible in life, as suggested by the fact that each ring overlaps its posterior successor (PI. 1, fig. 2) and by modalities of preservation (PI. 8, fig. 2). Mid-tail (Text-figs 3, 10a, 17; PI. 1, fig. 2; PI. 4, fig. 3; PI. 5, fig. 1; PI. 6, figs 1, 5-6; PI. 7, fig. 9; PI. 8, figs 1, 4, 9). The mid-tail skeleton consists of a dorsal element, the styloid, and paired ventral plates. The number of ventral plates is uncertain. Two lumps visible below the right ventral margin of the styloid of a specimen (PI. 7, fig. 9) are perhaps remnants of two right plates. The anterior region of the styloid decreases anteriorly in height (PI. 7, fig. 9) and sends out a process (ap; Text-fig. 17) which fits into the distal part of the fore-tail lumen (PI. 4, fig. 3; PI. 6, fig. 5; PI. 8, fig. 4). This process shows a deep antero-ventral excavation (he; Text-fig. 17; PI. 5, fig. 1 ; PI. 8, fig. 9), presumably for muscle insertion. The styloid bears two transverse blades (asb and psb; Text-fig. 17). The smaller anterior blade is nearly semicircular in outline viewed from the exterior. Its anterior face is flat and slightly inclined anteriorly and downward and bears five or six radial thickenings (rt; Text-fig. 17) near its outer margin. A depression (od; Text-fig. 17; PI. 5, fig. I ) is present between each thickening. The smooth posterior surface of the blade has a concave profile in lateral view and is transversely convex (PI. 7, fig. 9). It slopes downward and rearward, merges into the dorsal styloid surface and continues ventro-laterally. The free margin of the anterior blade is blunt. Its thick ventro-lateral ends turn inward and slightly downward, merging into the lateral styloid surfaces. The posterior blade is about twice as tall and wide as the anterior blade, has a parabolic outline in anterior view and is set lower than the anterior blade. Its anterior and posterior surfaces are mostly parallel to each other but diverge ventrally, so that the blade thickens in its lower third. The anterior surface is slightly concave in lateral view and joins the dorsal styloid surface vertically. The posterior surface is convex in lateral view and slightly depressed in its lower half. The dorsal part of the free margin of the posterior blade is sharp. Its ventro- lateral angles are thicker than the free margin and join the lateral styloid surfaces at a right angle (Text-fig. 17; PI. 8, fig. 9). The antero-ventral styloid excavation (PI. 5, fig. 1; PI. 8, figs 1, 9) is smooth and roughly semicircular in anterior view. Its central area is occupied by a semi-oval pit (sp; Text-fig. 17), posterior to which is a longitudinal groove (fig; Text-fig. 17). The groove and the pit can be compared with similar structures in other mitrates. In Eumitrocystella savilli (Beisswenger 1994, figs 3g, 8a-b), two 'horns’ project upward into the styloid cusps both from the anterior excavation and from the posterior interossicular cavity and represent the infilling of deep pits in the skeleton. Analogous 'horns’ are carried by the posterior interossicular cavities of the first few hind-tail ossicles. These structures were interpreted as resorption canals by Beisswenger, although, in fact, there is no obvious reason for thinking that they resulted from resorption. Similar horn-shaped canals project from the posterior interossicular cavities of the tail ossicles in Ateleocystites guttenbergensis (Kolata and Jollie, 1982). A pit in the EXPLANATION OF PLATE 3 Figs 1-4. Placocystella africana (Reed, 1925). Specimens in figs 1-3 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 4 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale. 1, latex cast of BMNH EE 5656 (JJS9), plate 6. 2, latex cast of BMNH EE 5648 (JJS1), almost complete external aspect of ventral head skeleton with the distribution of riblets and ribs and partial ventral surface of h, D7 and D8; dorsal cup of right pyriform body partially preserved. 3, latex cast of BMNH EE 5651 (JJS4), ventral ornament, ventral fore-tail plates and right oral spine. 4, latex cast of SAM 0011, ventral head skeleton and articulation facets of the oral spines. All x 5. PLATE 3 RUT A and THERON, Placocystella 218 PALAEONTOLOGY, VOLUME 40 antero-ventral excavation of the styloid of Mitrocystites mitra and Mitrocystella incipiens (Jefferies, pers. comm.) is in the same position as the semi-oval pit of P. africana. Behind the antero-ventral excavation, the ventral styloid surface is vaulted and delimited laterally by two stout, vertical processes (vlp; Text-fig. 17). Each process has a lateral face, merging into the lateral walls of the styloid, and a median face, which joins the vaulted ventral styloid surface. The lateral and median faces of each process meet along a blunt ventral keel (vk; Text-fig. 17). The left and right keels diverge outward posteriorly. The posterior surfaces of the processes are badly preserved. Their anterior surfaces bear a small, oval shallow area (osa; Text-fig. 17). The processes contacted the mid-tail plates in life, but the nature of this contact cannot be reconstructed. The ventral styloid surface is shown in Text-figure 1 7 ; it carries a poorly preserved longitudinal groove which probably housed the notochord/nerve chord complex in life. A pair of pits is visible on both sides of this groove. Each pit sends out a median sulcus perpendicular to the groove. The pits and the sulci are considered as traces of spinal ganglia and spinal nerves respectively (Jefferies 1986). Hind-tail (Text-figs 3, 7a, 18-19; PI. 7, fig. 9; PI. 8, figs 2-3, 5-9). Articulated tails are rare, but isolated ossicles are common. The highest recorded number of hind-tail segments is 35. Each segment consists of a dorsal ossicle and two ventral plates. The plates (PI. 7, fig. 9) are rectangular in lateral view and convex outward. The ventral hind-tail surface is rounded in cross section. The two plates of a segment meet mid-ventrally along a flat suture, ensuring a certain degree of rigidity. This rigidity is also suggested by the fact that ossicles and plates show a coarse ornament, presumably for a gripping action in life (see Jefferies 1984). The plates bear thick, transverse ridges, the shape and number of which vary in different plates, and a keel below their dorsal margins. The anterior and posterior plate margins are sinuous. The antero-dorsal angle of each plate contacts the postero- ventral angle of the preceding ossicle. The most anterior ossicle is slightly larger than the others and its posterior margin is recurved in lateral view (PI. 7, fig. 9). It also shows faint striations dorso-laterally. The second ossicle shows both striations and small knobs. All more posterior ossicles have knobs on their lateral surfaces, but no striations. The knobs (kn; Text- fig. 19a) vary in number, size and distribution in different ossicles and, like the ventral plate ridges, they presumably gripped sediment in life. The ossicles are triangular in cross section. Their lateral faces are gently convex outward and meet along a mid-dorsal edge (mde; Text-figs 18a, 19a) which slopes forward and downward. Its posterior end is the dorsal apex (da; Text-figs 18, 19a). Each ossicle (PI. 8, fig. 5) has an anterior, a posterior, a ventral and two lateral surfaces. The dorsal half of the anterior surface is occupied by an oval depression (aod; Text-fig. 18a), delimited ventro-laterally by two ascending ridges (ar; Text-fig. 18a). These ridges approach the median plane of the ossicle ventrally, but do not contact each other. They diverge dorso-laterally and become confluent with the central part of the thick anterior ossicular margins (aom; Text-fig. 18a). The oval depression is delimited dorso-laterally by the dorsalmost parts of the two anterior ossicular margins, merging into the anterior end of the mid-dorsal edge. The dorsal third of the oval depression is much deeper than the ventral two-thirds and is cone-shaped (ce ; Text- fig. 18a). The deepest part of it is occupied by a small circular pit. The ventral two-thirds of the oval depression explanation of plate 4 Figs 1-6. Placocystella africana (Reed, 1925). Specimen in fig. 1 from locality D, Gydopas; Ceres Subgroup, Gydo Shale; specimens in figs 2-4 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 5 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 6 from Tunnel Siding, De Dooms; Ceres Subgroup, Voorstehoek Shale. 1, latex cast of SAM 0061, isolated oral spine and V9. 2, latex cast of BMNH EE 5662 (JJS1 5), central part of the postero-dorsal head skeleton; note the different shapes of the anterior margins of h and i. 3-4, latex casts of part and counterpart of SAM 0114: 3, partial dorsal aspect of the head, anterior margin of lower lip, spine articulations, fore-tail rings surrounding the styloid and anterior styloid blade; 4, partial ventral aspect of the head; note the impression left by the partial course of the oblique ridge; right oral spine visible in ventral view. 5, latex cast of SAM 0106, partial anterior area of ventral head steinkern and stereom structure of the oral spines. 6, latex cast of BMNH EE 5668 (JJS21); note the articulation of the upper lip plates and the optic foramen delimited by the hypocerebral processes. All x 5. PLATE 4 RUT A and THE RON, Placocy Stella 220 PALAEONTOLOGY, VOLUME 40 text-fig. 10. Placocystella africana (Reed, 1925). a-b, natural external mould of dorsal and ventral head skeleton of SAM 0104; locality D, Gydopas; Ceres Subgroup, Gydo Shale; distribution of ribs and riblets; both x 5. houses a shallow, anterior interossicular groove (aig; Text-fig. 18a), flanked on both sides by a thin, vertical ridge (vr; Text-fig. 18a). The latter becomes fainter dorsally and ventrally before disappearing. The ventral half of the anterior ossicular surface bears two shallow triangular facets (tf; Text-fig. 18a), delimited dorso-laterally by the anterior ossicular margins, dorso-medially by the ascending ridges and ventrally by the upper surface of two articulation bosses (see below). Each facet sends out a ventro-lateral groove which joins a semi-elliptical depression (sd; Text-figs 18a, 19b). Each articulation boss (ab; Text-figs 18a, 19b) is shaped like an inverted ‘ L’, the vertical arm of which is concave laterally and forms a continuous surface with the above-mentioned semi-elliptical depression. The median surface of the lower arm contributes to the mid-ventral groove (see below). The horizontal arm of each boss shows: a flat upper face joining the triangular facet posteriorly; a sausage-shaped median face contributing to the foremost part of the lateral wall of the mid-ventral groove; a concave ventral face merging into the dorsal part of the semi-elliptical depression; and a lateral face, separated from the lower third of the anterior ossicular margin by the short groove connecting the triangular facet to the semi-elliptical depression. RUTA AND THERON: DEVONIAN MITRATES 221 The ossicular mid-ventral groove (clg; Text-fig. 19b) probably housed the notochord in life. A poorly preserved, ribbon-like structure visible along the groove is interpreted tentatively as a trace of the dorsal nerve chord. Two narrower grooves project laterally from the ribbon-like structure and end in two pits. Grooves and pits are interpreted tentatively as traces of nerves and ganglia (tsg and tsn; Text-fig. 19b) respectively. The posterior ossicular surface bears two triangular platforms (tp; Text-fig. 18b) that fit into the triangular depressions of the anterior face of the successive posterior ossicle. A posterior interossicular groove (pig; Text- fig. 18b) is visible in the middle of a posterior oval depression (pod; Text-fig. 18b). Ventro-lateral to each triangular platform is a rounded area (rd; Text-fig. 18b), which accommodates the dorsal arm of the articulation boss of the succeeding ossicle. The uppermost part of the posterior ossicular surface is flat (pbs; Text-fig. 18b). A protuberance (pp; Text-fig. 18b) on the lower third of the posterior margin fits into a notch (dn; Text-fig. 18a) on the anterior margin of the succeeding ossicle (PI. 7, fig. 9; PI. 8, figs 3, 5-6, 8). The ventral ossicular margins are triangular in cross section and are articulated with the dorsal margins of the hind-tail plates. The rearmost part of the ventral ossicular margin bears a facet for the articulation of the successive posterior ventral plate (afvp; Text-figs 18b, 19). Internal anatomy of the head Dorsal steinkern (Text-figs 12a, 13a, 14, 15a). The dorsal steinkern carries an oblique groove which runs from its anterior right to its posterior left and divides it into two unequal areas. The groove is asymmetrical in cross section along most of its length, being steeper to the right than to the left. The posterior two-thirds of the groove (ppog; Text-fig. 21a; Text-figs 13-14, 15a), almost straight and asymmetrical in cross section, runs from the posterior left angle of the head to a mid-dorsal pit (mdp; Text-fig. 21a). Before reaching it, the groove deepens and widens. The anterior third (apog; Text-fig. 21a) of the groove runs from the mid-dorsal pit to the right anterior angle of the upper lip. In front of the pit the groove turns sharply rightward and becomes shallower and more symmetrical in section. Before joining the right angle of the upper lip, the groove becomes convex rightward and somewhat deeper again. Detailed comparisons between mitrates and extant chordates (especially tunicates) are crucial for the interpretation of the internal anatomy of the mitrate head. According to Jefferies (1986), the oblique groove marks the right boundary of the region where the left anterior coelom (left mandibular somite of vertebrates) was in contact with the dorsal head skeleton. The left anterior coelom would overlie the left pharynx. An elongate band (dl; Text-fig. 21a) runs obliquely and posteriorly from the mid-dorsal pit (Text-fig. 13a). It presumably represents the impression of the cavity of the dorsal lamina (Jefferies 1986, pp. 93, 270, 279), a fold of tissue which occupies the mid-dorsal line of the pharynx in some tunicates. In Mitrocystella incipiens, the position of the dorsal lamina is marked by two parallel lines on the dorsal steinkern (Jefferies and Lewis 1978; Chauvel 1981 ; Jefferies 1986), whereas in P. africana these lines are irregular. A faint line on the steinkern of h (Text-fig. 13a) would correspond to the left boundary of the right pharynx (lbrp; Text-fig. 21a). Following Jefferies (1986), the area of the dorsal steinkern right of the impression left by the dorsal lamina corresponds in position to the rightmost part of the right anterior coelom (right mandibular somite of vertebrates), which would overlie the right pharynx in life. Most of the viscera were probably housed below the triangular area comprised between the posterior two-thirds of the oblique groove and the left boundary of the right pharynx. Two small grooves (rag and rpg; Text-fig. 21a; see also Text-fig. 13a) are visible to the right of and anterior to the right pyriform body in one individual. They may represent a partial right peripheral canal. In other mitrates (e. g. Mitrocystites mitra ), the left and right peripheral canals are visible along the inner sides of the marginal dorsal plates and would have housed the nerves labelled as n2 (see below). Another groove is visible in the left postero-lateral region of the dorsal steinkern (lg; Text-fig. 21a). Anterior to this groove, a short ridge (r; Text-fig. 21a) bends downward and slightly medianward, running along the posterior surface of the head steinkern. Comparison with M. incipiens suggests that this ridge may represent part of the rectum. The specimen in Text-figure 13a shows the position of the ascending part of the rectum, left of the rearmost part of the oblique groove. The rectum would open into a left atrium, according to Jefferies (1986; see also Jefferies and Lewis 1978). However, the position of the anterior boundary of the left atrium in P. africana is unknown. A bump in front of the rag and rpg grooves (see Text-fig. 21a) is interpreted tentatively as a trace of the anterior boundary of the right atrium (abra; Text-fig. 21a). A crescent-shaped line in front of the posterior margin of the dorsal steinkern (abpc; Text-fig. 21a; Text- figs 13, 14a) corresponds to the anterior wall of the posterior coelom. According to Jefferies (1986), the latter would be homologous with the tunicate epicardia. PALAEONTOLOGY, VOLUME 40 222 text-fig. 11. Placocystella africana (Reed, 1925). a, latex cast of SAM 0101; locality K, Klipfontein; Ceres Subgroup, Voorstehoek Shale; partial dorsal head skeleton, partial dorsal aspect of the ventral head skeleton, endostylar trace, oral spine insertions and anterior margin of ventral head skeleton; x5; b, interpretative sketch. EXPLANATION OF PLATE 5 Figs 1-6. Placocystella africana (Reed, 1925). Specimen in fig. 1 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimens in figs 2, 4 and 6 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale ; specimen in fig. 5 from locality K, Klipfontein ; Ceres Subgroup, Voorstehoek Shale ; specimen in fig. 3 from locality D, Gydopas; Ceres Subgroup, Gydo Shale. 1, latex cast of SAM 0112, anterior aspect of the styloid; note the radial thickenings on the anterior blade and the pit on the dorsal surface of the antero- ventral excavation; x 12. 2, SAM 0107, natural external mould of V12; the impression left by the endostylar trace is visible; x 5. 3, latex cast of SAM 0061, a young individual showing a faint suture between V12 and p; note the distribution of ribs and riblets and the transverse keel on e; x 5. 4, latex cast of SAM 0013, with e, p, 9 and VI 2 articulated; riblets and concentric growth lines are visible; x 4. 5, latex cast of SAM 0092, showing the differences in the kind and distribution of ornament on the posterior third of the ventral head skeleton and growth lines; note the finely granular stereom texture of the external surface of VI 2; x 6. 6, latex cast of SAM 0 1 06, showing the morphology of the right oral spine, its articulation with d, the anterior margin of V6 and the depression lying to the right of the spine articulation on the dorsal surface of V6; x 12. PLATE 5 RUTA and THERON, Placocystella 224 PALAEONTOLOGY, VOLUME 40 text-fig. 12. Placocystella africana (Reed, 1925). Reconstruction of the internal aspect of the head. A, dorsal view; b, ventral view; c, posterior view; d, anterior view (soft parts of the head removed; left pyriform body drawn in part to show the left acoustic ganglion). The enlarged, deep part of the oblique groove behind the mid-dorsal pit probably represents, by comparison with Placocystites forbesianus, the site of the ciliated organ (opening of the neural gland duct). The pit is the point from which a mid-dorsal process (part of which is visible as a short stump in Text-fig. 13b) would extend inward and rearward to stiffen the free edge of the dorsal lamina (Jefferies and Lewis 1978; Jefferies 1986). The anteriormost region of the dorsal steinkern shows two bumps (loc and roc; Text-fig. 21a) underneath plates b and c. These are interpreted as olfactory cups on the basis of their position inside the mouth. A faint transverse groove just in front of the posterior margin of n (Text-fig. 13a) may represent the impression of the velum (vg; Text-fig. 21a). The likely position of the head chambers is shown in Text-figure 21a. Ventral steinkern (Text-fig. 12b; PI. 7, fig. 1). A groove runs from in front of the p/V12 suture to a point slightly right of the plane of bilateral symmetry. If the interpretation of the internal anatomy of the mitrate head and the dorso-ventral orientation of P. africana are correct, it is reasonable to interpret this groove as a trace of the skeletal support for the endostyle (et; Text-fig. 21a), for the groove is mid-ventral in position and the only RUTA AND THERON: DEVONIAN MITRATES 225 text-fig. 13. Placocystella africana (Reed, 1925). a, partial dorsal head steinkern of BMNH EE 5649 (JJS2); Swaarmoed Pass; Ceres Subgroup, Voorstehoek Shale; oblique groove, mid-dorsal pit, velar groove, foremost part of dorsal lamina, right boundary of left pharynx, anterior boundary of right atrium, right olfactory organ and partial course of the right nerve n2. b, latex cast of the same specimen, showing the position of the rectum and of the anterior boundary of the left atrium; both x 5. important mid-ventral organ in the pharynx of primitive living chordates is the endostylar gland (Jefferies 1986). The endostylar trace is V-shaped in cross section and slightly convex leftward in most specimens, although variation exists (Text-fig. 11). It stops posteriorly in front of a transverse excavation extending between the pyriform bodies (see below) and is flanked on both sides by a fainter, lateral groove (PI. 7, fig. 1 ). The latter runs parallel to the endostylar groove for most of its length, but diverges outward posteriorly. Another faint line, visible on both sides of the endostylar trace (PI. 7, fig. 1), runs parallel to the endostylar and lateral groove, but is shorter. The steinkern of VI 2 has a dappled aspect and shows concentric growth lines. Allanicytidium flemingi from New Zealand (Caster and Gill 1967, fig. 361 ; Caster 1983, fig. 1) shows a similar presumed ventral endostylar trace (er; Text-fig. 22). This is concave leftward and runs from in front of the p/V12 suture to slightly right of the centre of VI 2, where it turns leftward. A depression on both sides of the endostylar ridge follows the curvature of the latter. The left depression (Id; Text-fig. 22) is deeper than the right one for most of its length, but becomes shallower anteriorly. Left of this depression, the dorsal surface of V12 shows two triangular areas (tsal and tsa2; Text-fig. 22) surrounded by a faint, tortuous crest (lfc; Text-fig. 22). The crest delimits three other depressions laterally (tsdl, tsd2 and tsd3: Text-fig. 22), continues anteriorly for a short distance flanking the endostylar ridge, and disappears before the latter changes direction. The depression on the right of the endostylar ridge (rd; Text-fig. 22) is flanked by another crest (rfc; Text-fig. 22). A transversely elongate pit (ep; Text-fig. 22) lies anterior to the endostylar ridge. The dappled aspect of VI 2 in P. africana and the depressions and ridges on both sides of the endostylar ridge in A. flemingi are difficult to interpret; possibilities are blood vessels, nerves, or viscera. By combining information from different specimens it is possible to detect the existence of three sub-circular depressions on the ventral steinkern of P. africana, corresponding to central buttons on the dorsal surface of 226 PALAEONTOLOGY, VOLUME 40 some ventral plates (Text-fig. 14; PI. 2, fig. 3). The likely existence of three other buttons is deduced from a comparison with A.flemingi. The buttons are remnants of the inner calcitic layer of the three-layered ventral skeleton typical of all mitrates. Their function is unknown. Two asymmetrical buttons in front of the endostylar trace in one specimen (PI. 7, fig. 7) resemble similar structures in Barrandeocarpus norvegicus (Craske and Jefferies 1989, figs 14b, 17; pi. 13, fig. 2). A depression on the anterior margin of V6 and V9 (Text-figs 9, 1 1 ; PI. 1, fig. 2; PI. 4, fig. 3; PI. 5, fig. 6) may represent the external opening of an inlet valve or a continuation of the olfactory organs. Nervous system According to the calcichordate interpretation of mitrates, the inflated structure at the head/tail junction is the brain, divided into prosencephalon and deuterencephalon (pros and deut; Text-fig. 16; PI. 4, fig. 6; PI. 7, figs 2-3). As in Placocystites forbesiamis, the prosencephalon of P. africana is lens-shaped, but with no obvious distinction between telencephalon and diencephalon. The deuterencephalon is less expanded transversely than in P. forbesianus. Two hypocerebral processes (hyp; Text-fig. 16), almost in contact medially, delimit a transverse optic foramen (of; Text-fig. 16; PI. 4, fig. 6; PI. 7, fig. 6). In one individual (PI. 7, fig. 6), a ridge projecting upward from the right half of the foramen probably represents the right cispharyngeal optic nerve. A left cispharyngeal optic nerve has not been observed. Above the foramen is a thick crescentic body (premandibular somite of vertebrates; Jefferies 1986, fig. 8.23). The postero-ventral head steinkern is reconstructed in Text-figure 20, based on specimen BMNH EE 5653 (JJS6; PI. 7, figs 1, 7). The groove extending laterally out to each of the two pyriform bodies corresponds to a skeletal wall behind the p/V12 suture. The upper edge of this wall is blunt. Its posterior surface is convex, whereas its anterior surface is concave. The latter marked the anterior boundary of the posterior coelom ventrally. Two ridges (nO; Text-fig. 20; PI. 7, fig. 1), visible on the anterior face of the transverse groove on the postero- ventral steinkern, are probably traces of the nerves nO, present in Placocystites forbesiamis , Barrandeocarpus norvegicus and Eumitrocystella savilli. These nerves are believed to have supplied the endostyle in life on the basis of their position on the internal surface of the ventral head skeleton (Jefferies and Lewis 1978; Jefferies 1986, p. 278). The fact that the nerves nO in P. africana are visible as ridges on the steinkern suggests that the transverse wall behind the endostylar ridge (PL 7, fig. 7) was formed in life entirely of the outer calcitic layer. In Mitrocvstella incipiens and P. forbesianus, this region is formed of the inner calcitic layer and the nerves run between the outer and the inner layers. The thickenings labelled as bpc in Text-figure 20 may represent the median, lower parts of the two palmar complexes. In other mitrates, these structures give rise to various nerves. With the exception of nerves n2, the other nervous branches are not observed in P. africana. Nerves n2 (Text-fig. 21a), regarded by Jefferies (1986) as homologous with the maxillary branches of the trigeminal nerves, are partially preserved as ridges on the lateral walls of the steinkern and terminate in the buccal cavity, lateral to the olfactory cups (Text-figs 13, 14b). The pyriform bodies (trigeminal ganglia in Jefferies’ view) are antero-lateral to the deuterencephalon (Text- figs 12, 13a, 15b, 20; PI. 7, fig. 1), but their median connections are unknown. Their major axis is inclined antero-medially. A small lump and a slender ridge visible near the left pyriform body are tentatively interpreted as a left auditory ganglion and a left acoustic nerve respectively (lag and lan; Text-fig. 20; PI, 7, fig. 1; see Cripps (1990) for the interpreatation of similar structures in the mitrate Chaubelia discoidalis). A small raised lump right of the right pyriform body (Text-figs 13, 15b) is tentatively interpreted as the right auditory ganglion. EXPLANATION OF PLATE 6 Figs 1-6. Placocystella africana (Reed, 1925). Specimens in figs 1-4 and 6 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 5 from locality A, Klipfontein ; Ceres Subgroup, Voorstehoek Shale. 1, 3, latex casts of part and counterpart of BMNH EE 5656 (JJS9): 1, fore-tail rings (left part of the photograph), styloid (centre) and hind-tail ossicles (right part of the photograph); 3, isolated ossicles (left part of the photograph). 2, 4, part and counterpart of BMNH EE 5662 (JJS15), fore-tail rings: 2, natural ventral external mould of part; 4, latex cast of dorsal extenal mould of counterpart. 5, latex cast of SAM 0011, dorsal head skeleton, oral spines arid fore-tail rings and partially disrupted styloid. 6, latex cast of SAM 0112, articulation facets of oral spine and anterior aspect of the styloid. All x 5. PLATE 6 RUTA and THERON, Placocystella 228 PALAEONTOLOGY, VOLUME 40 text-fig. 14. Placocystella africana (Reed, 1925). A, partial dorsal and ventral head steinkerns of SAM 0009; locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; anterior boundary of posterior coelom, rearmost part of oblique groove and left boundary of right pharynx; a depression is visible on the steinkern of V6; x 6. b, partial dorsal and ventral head steinkerns of BMNH EE 5654 (JJS7); Swaarmoed Pass; Ceres Subgroup, Voorstehoek Shale; a depression on the steinkern of V9 and the partial course of the left nerve n2 are visible; external moulds of the oral spines preserved; x 6. Family Incer tae sedis Genus bokkeveldia gen. nov. Derivation of name. From the Bokkeveld Group of South Africa. Type species. Bokkeveldia oosthuizeni sp. nov., the only known species. Diagnosis. Five transverse rows of ventral plates; first, third and fourth row with five plates each; second and fifth row with four and three plates respectively; VI 1 and V13 smaller than V7, V8, V16 and V18; V17 contacts p, V12, V16 and V18; V12 separated from V3 by V7 and V8; e separated from 6 by p; ribs on V15, V16, V18, V19, e, p and 9. RUTA AND THERON: DEVONIAN MITRATES 229 text-fig. 15. Placocystella africana (Reed, 1925). A, partial dorsal head steinkern of SAM 0110; locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; oblique groove, mid-dorsal pit and right olfactory organ; the natural external moulds of some ventral fore-tail plates and of the oral spines are also preserved; x 5. b, incomplete dorsal head steinkern of SAM 0109; locality and horizon same as a; the right pyriform body is visible; x 5. Bokkeveldia oosthuizeni gen. et sp. nov. Text-figure 6a-b 1984 Placocystella sp. (a) Rennie; Oosthuizen, p. 134, table 3. Derivation of name. After Mr R. D. F. Oosthuizen of Zwartskraal, Cape Province, discoverer of the holotype, for his contributions to the biostratigraphy and palaeontology of the Bokkeveld Group. Material. Complete external mould of a ventral head skeleton slightly disrupted in the posterior part. The specimen (Oosthuizen’s collection, no. 148) was legally bequeathed to the South African Museum, Cape Town. Type locality and horizon. Gamka Poort, Prince Albert Area, Gydo Shale; 33°18'S, 21°38'E. Diagnosis. As for the genus, by monotypy. 230 PALAEONTOLOGY, VOLUME 40 Remarks. Although known from poor material, Bokkeveldia oosthuizeni shows a remarkable morphology. Its ventral plate arrangement is here chosen as a reference to identify homologous plates in all the mitrates with a standardized ventral head skeleton. In the vast majority of the anomalocystitids, the ventral head skeleton consists of a fixed number of plates arranged in transverse rows. The plates can be subdivided topographically in mid-ventral, admedian ventral and lateral ventral elements. A comparison with Bokkeveldia makes it possible to assign a particular plate to one or the other of these three spatial groups. The identification of homologous plates in the anomalocystitids will be dealt with extensively elsewhere. Bokkeveldia oosthuizeni is similar to an unnamed lower Devonian mitrate from Morocco described and figured by Regnault and Chauvel (1987, fig. 1) and known from a single individual (specimen IGR 16639, Institut de Geologie, University of Rennes, France) of which the head steinkern and part of the dorsal skeleton are known. The ventral steinkern of the Moroccan mitrate bears the impressions of polygonal plates arranged in at least four transverse rows, but the incomplete preservation of this fossil makes it difficult to draw more accurate comparisons with Bokkeveldia. PHYLOGENETIC ANALYSIS Thirteen taxa were chosen for this study: Allanicytidium flemingi Caster and Gill (Reefton Mudstone, New Zealand), Ateleocystites guttenbergensis Kolata and Jollie (Guttenberg Formation, Wisconsin, USA), Australocystis langei Caster (Ponta Grossa Shale, Brazil), Barrandeocarpus jaekeli Ubaghs (Letna Formation, Czech Republic), B. norvegicus Craske and Jefferies (Langoyene Sandstone, Norway), Eumitrocy Stella savilli Beisswenger (Ouine-Inirne Formation, Morocco), Mitrocystella barrandei Jaekel (Sarka Formation, Czech Republic), M. incipiens (Barrande) (Dobrotiva Formation, Czech Republic, and Formation de Traveusot, France), Notocarpos garratti Philip (Humevale Formation, Victoria, Australia), Placocystella africana (Reed) (Gydo and Voorstehoek shales, South Africa), Placocystites forbesianus de Koninck (Wenlock Limestone, England), Tasmanicytidium burretti Caster (Richea Siltstone, Tasmania, Australia) and Victoriacystis wilkinsi Gill and Caster (Dargile Beds, Victoria, Australia). The morphological data are gleaned mainly from the works of Caster (1954, 1983), Gill and Caster (1960), Caster and Gill (1967), Ubaghs (1967, 1979), Jefferies and Lewis (1978), Chauvel (1981), Philip (1981), Kolata and Jollie (1982), Jefferies (1986), Craske and Jefferies (1989), Parsley (1991) and Beisswenger (1994). All characters were left unweighted and unordered and processed with PAUP 3.1.1 (Swofford 1993) under the ACCTRAN optimization, which explains homoplasy in terms of reversals. One EXPLANATION OF PLATE 7 Figs 1-9. Placocystella africana (Reed, 1925). Specimens in figs 1-3, 7 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimens in figs 5-6 from locality K, Klipfontein; Ceres Subgroup, Voorstehoek Shale; specimens in figs 4 and 8 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 9 from locality J, Lakenvalleidam; Ceres Subgroup, Voorstehoek Shale. 1, 7, BMNH EE 5653 (JJS6) and latex cast; left pyriform body, left acoustic nerve and left auditory ganglion; note the endostylar groove, flanked on both sides by two fainter grooves, and the dappled aspect of the steinkern of V12; growth lines and two asymmetrical ‘buttons’ visible in front of the endostylar trace; x 5. 2-3, BMNH EE 5654 (JJS7), showing the brain: 2, dorsal view; 3, anterior view; x 5. 4. latex cast of SAM 0013, endostylar trace and partially eroded lower cups of pyriform bodies; x 5. 5, latex cast of SAM 0132, endostylar ridge and adjacent ridges; x 6. 6, SAM 0050, head steinkern in posterior view, optic foramen and right cyspharyngeal optic nerve running laterally from the right half of the foramen; x 5. 8, SAM 0109, rearmost part of internal surface of dorsal head skeleton with the impression of the anterior boundary of the posterior coelom and part of the brain; x 5. 9, latex cast of SAM 0139, mid- and hind-tail in right lateral view; x 7. PLATE 7 RUT A and THERON, Placocystellci 232 PALAEONTOLOGY, VOLUME 40 parsimonious tree (branch-and-bound algorithm; length = 51; Cl = 0 857 excluding uninformative characters nos 27 and 35; RI = 0-929) was found. Mitrocystella barrandei and M. incipiens were used as outgroups. In Austra/ocystis, the multi-state coding for character no. 16 (2 or 3) means uncertainty. The analysis was subject to bootstrap, which estimates the support of a hypothesis of phylogeny (in our case based on parsimony) through repeated sampling of characters, regardless of the historical reality of a clade (Hillis and Bull 1993). The 50 per cent, majority-rule consensus tree (Text-fig. 25; 1000 replicates; branch-and-bound search) shows high bootstrap values at nodes D (100 per cent.), I (100 per cent.) and E (97 per cent.), intermediate values at F (88 per cent.), C (85 per cent.) and J (83 per cent.) and low values at H (70 per cent.), L (63 per cent.) and K (50 per cent.). Node F is supported by three character-state changes: 9 refers to the regular ventral plating pattern; 22 refers to the presence of oral spines (their insertions are visible in Ateleocystites and Allanicytidium ); 16 refers to the differentiation of V12 which in Ateleocystites occupies the second ventral row and extends anteriorly and posteriorly to separate in part the admedian plates of the first and third row respectively. Node G is supported by character-state changes nos 13, 16 and 24. Character no. 13 concerns the absence of admedian elements in the third ventral row. Reversal is found in Victoriacystis, in which two small plates seem to be present postero-lateral to the large mid-ventral plate (Gill and Caster 1960, fig. lib; pi. 9, fig. 2; pi. 10, fig. 2). These elements, VI 1 and VI 3, are also small in Bokkeveldia by comparison with V10 and VI 4. If the identification of VI 1 and VI 3 in Victoriacystis is correct, their size may be interpreted as a stage of their progressive reduction in passing from the less derived anomalocystitids to the allanicytidiids, which lack VI 1 and VI 3. The absence of VI 1 and VI 3 in Placocvstites may also be interpreted as a parallelism with the Allanicytidiidae (this is evident when the DEFTRAN optimization is in effect). Characters nos 16 and 24 concern the expansion of V12 and the absence of Dll. Node H is supported by character-state changes nos 12 (second ventral row formed by two lateral elements), 30 (flexible upper lip) and 31 (lack of oral platelets). The second ventral row is absent in Placocvstites , as tentatively deduced from the fact that in this mitrate three elements form the anterior row, of which one is mid-ventral. On the basis of their relative position with respect to other plates, these elements are probably homologous with V2-V4 of Victoriacystis and Bokkeveldia. The monophyly of the Allanicytidiidae (node I) is supported by ten state changes: 7, 10-11, 14-16 and 23 refer to the arrangement of the ventral plates (preserved only in part in Austra/ocystis langei ); 3 refers to the styloid blades (poorly preserved in Austra/ocystis; unknown in Tasmanicytidium burretti ); 29 and 32 refer to plates b, n and c. The distribution of dorsal ribs, the general head proportions and the shape of D10 and D12 are similar in Notocarpos garratti and in Placocystites forbesianus. Node J is supported by the presence of ventral riblets (21; unknown state at node F) and by the rearward extension of D10 (33). P. africana is the sister taxon of Allanicytidium plus Australocvstis (node K; character no. 37, presence of dorsal riblets). Allanicytidium and Australocystis (node F) share the similar shape of D10 and D12 (34). EXPLANATION OF PLATE 8 Figs 1-9. Placocystella africana (Reed, 1925). Specimens in figs 1, 3-5 and 7 from Swaarmoed Pass, Ceres; Ceres Subgroup, Voorstehoek Shale; specimen in fig. 2 from locality A, Klipfontein; Ceres Subgroup, Voorstehoek Shale; specimens in figs 6 and 8 from locality D, Gydopas; Ceres Subgroup, Gydo Shale. 1, 9 BMNH EE 5662 (JJS15): 1, natural mould of styloid; x 10; 9, latex cast of styloid in antero-lateral view; x 7. 2, latex cast of SAM 0014, latero-ventral aspect of the head; the hind-tail is bent under the head; x 5. 3, latex cast of BMNH EE 5658 (JJS1 1), isolated hind-tail ossicles; x 5. 4, latex cast of of SAM 01 14, fore-tail rings surrounding the anterior styloid process; x 5. 5, latex cast of BMNH EE 5656 (JJS9), isolated hind-tail ossicle in anterior view; x 12. 6, 8, part and counterpart of SAM 0104, external mould of the hind-tail; note the ossicular articulations; x 5. 7, specimen BMNH EE 5662 (JJS15), natural mould of posterior ossicular surface; x 10. PLATE 8 RUTA and THERON, Placocystella 234 PALAEONTOLOGY, VOLUME 40 plate D1 plate i text-fig. 16. Placocystella africana (Reed, 1925). Reconstruction of the posterior head skeleton, show- ing the two hypocerebral processes, the optic foramen and the division of the cerebral basin into a prosencephalar and a deuterencephalar part. Abbrevi- ations as in the text. text-fig. 17. Placocystella africana (Reed, 1925). Reconstruction of the styloid in antero-ventral view. Abbreviations as in the text. text-fig. 18. Placocystella africana (Reed, 1925). Reconstruction of a tail ossicle, a, anterior view; b, posterior view. Abbreviations as in the text. RUTA AND THERON: DEVONIAN MITRATES 235 text-fig. 19. Placocystella africana (Reed, 1925). Reconstruction of a tail ossicle, a, left view; b, ventral view. Abbreviations as in the text. mde da af vp A text-fig. 20. Placocystella africana (Reed, 1925). Reconstruction of the postero-ventral head steinkern. Abbreviations as in the text. Six minimal trees were found when Bokkeveldia was included in the analysis. Their strict consensus placed this mitrate with Ateleocystites , Placocystites and Victoriacystis in a polytomy. The monophyly of the Allanicytidiidae was supported in all trees and the relative positions of the other ingroup taxa did not change. DISCUSSION The allanicytidiids show a remarkable external bilateral symmetry, also evident in the plate arrangement, and a peculiar ornament. Jefferies (1984, p. 308) showed that in the ontogeny of Placocystites forbesianus the first ribs appeared ventrally and tended to be short and crescentic, but became elongated, joined together, and straightened as the head grew; at any given stage, therefore, the straightest and most continuous ribs are found in the posterior part of the ribbed area ’. The pattern of ornament in the allanicytidiids suggests that, during their evolutionary history, they acquired riblets in the same ontogenetic order of appearance of ribs as observed in P. forbesianus. This sequence seems to reflect the phylogenetic order of appearance of ribs. Mitrocystites mitra shows few, short ribs along the postero-lateral head margins. Mitrocystella incipiens bears postero-ventral ribs. Barrandeocarpus spp. and many anomalocystitids possess transverse dorsal ribs. In Notocarpos , ventral ribs are confined to the postero-lateral areas of the head floor and become broken medially (Philip 1981, fig. 5c). Tasmanicytidium has a smooth dorsal skeleton, but the ventral skeleton is covered with few riblets and the lateral head walls show broken ribs. In P. africana , dorsal riblets replace most of the ribs. The few, sinuous dorsal ribs of P. africana 236 PALAEONTOLOGY, VOLUME 40 text-fig. 21. Placocystella africana (Reed, 1925). Reconstruction of the chambers of the head, a, dorsal aspect; b, cross section behind the widest point of the head. Abbreviations: AC, patent part of right anterior coelom; BC, buccal cavity; LA, left atrium; LP, left pharynx; PC, posterior coelom; RA, right atrium; RP, right pharynx. Other abbreviations as in the text. In B the continuous line surrounding the left pharynx is the virtual left anterior coelom; the dotted line surrounding the right pharynx is the virtual part of the right anterior coelom. The peripharyngeal bands (lppb and rppb) are hypothetical. • • • • • • •••••••• seem to derive from the confluence of riblets. The ventral riblets are more numerous in Placocystella than in Tasmanicytidium. Allanicytidium has numerous dorsal riblets. Few ribs are present near the postero-lateral angles of its head roof. Its ventral ornament is unknown. Changes in the ornament probably occurred through the retention of a juvenile feature (riblets) found in less derived forms. Australocystis, Victoriacystis and Tasmanicytidium deserve a final comment. Australocystis is certainly an allanicytidiid. The holotype (Caster 1954, fig. 1 ; pi. 8, figs 1-2) shows a square dorsal elevation, resulting from the post-mortem squashing of the dorsal skeleton against V12 (a condition observed also in some specimens of P. africana). V6 and V9 are visible in dorsal view and their anterior margin is asymmetrical in section, as in all other allanicytidiids. Furthermore, the dorsal plating pattern of Australocystis is very similar to that of Allanicytidium. In Victoriacystis , a large mid-ventral plate in front of V17 probably contacts anteriorly V2-V4 and may correspond to V12, which incorporated or eliminated V7 and V8. V6 and V9 show the same relative position with respect to the surrounding plates as in Ateleocystites and Bokkeveldia. The third, fourth and fifth rows in Victoriacystis are comparable to those of Bokkeveldia , except for the size of V12. The transition from V. wilkinsi to N. garratti was probably marked by the loss of V1-V5, VI 1, VI 3, and V16-V18. Our interpretation of the anatomy of Victoriacystis rests on the RUTA AND THERON: DEVONIAN MITRATES 237 text-fig. 22. Dorsal aspect of plate V12 in Allani- cytidium flemingi. Abbreviations explained in the text (redrawn after Caster and Gill 1967). text-fig. 23. Cladogram of a selected number of anomalocystitid and mitrocystitid mitrates. Syna- pomorphies are discussed in the text. © (0 p P o 'c p < 12 (0— >1 ); 30 (0— >1); 31 (0— >1 ) G 13 (0— >1 ); 16 (1 — >2); 24 (0~>1 ) F 9 (0— >1 ); 16 (0— >1 ); 22 (0— >1 ) 8 (0— >1 ); 15 (0— >1); 18 (0~>1); 25 (0 — >1 ); 26 (1— >2) D 2 (0~>1 ); 4(0— >1 ); 7 (0— >1); 1 1 (0— >1); 19 (0— >1 );20 (0~>2); 26 (0~>1); 28 (0— >1 ); 36 (0-3-1) 1 (0— >1); 5 (0— >1); 6 (0->1); 38 (0— >1 ) A i! i- w o o « » o s —i ^ —i — ;> {2 a n n> Vs S Q. ^ . £ a. o s5?- ^ o 5. I?! 3 o- • O o-t3 text-fig. 24. Interrelationships of the Allani- cytidiidae. Characters discussed in the text. 238 PALAEONTOLOGY, VOLUME 40 O) $ -5 V) O ,1 £ JO § o' o o E - ■S « e g> § I E .w .3 O) S O' 34 (0->1) 37 (0~>1) 21 (0— >2); 33 (0— >1 ) 3 (0— >1); 7 (1— >2); 10 (0— >1); 11 (1— >0); 14 (0— >1 ); 15 (1~>0); 16 (2— >3); 23 (0— >1); 29 (0->1); 32 (0->1) text-fig. 24. Interrelationships of the Allani- cytidiidae. Characters discussed in the text. S' E UJ tr vj E $ •E O E g> ^ I £ .S5 o5 S o' 63% 50% 83% 100% 70% 77% 88% 97% 100% 85 % text-fig. 25. Bootstrap 50 per cent, majority-rule consensus tree. original illustrations in Gill and Caster (1960), and further work on the morphology of this mitrate is necessary. For Caster (1983, fig. 4), Tasmanicytidium had a single centro-dorsal plate and the matrix infilling lying to the right of it represented the area formerly occupied by the admedian half of a marginal plate. Flowever, the marginal plates of this mitrate show no signs of breakage. Perhaps a second centro-dorsal was present in life to the right of the preserved one, suggesting that D10 was wider than D12. An accurate reconstruction of Tasmanicytidium depends upon more complete finds. Acknowledgements. We thank Dr R. P. S. Jefferies (NHM Palaeontology Department, London), who suggested and revised this work, and an anonymous referee for constructive criticism. MR thanks Drs RUTA AND THERON: DEVONIAN MITRATES 239 L. R. M. Cocks, Keeper, and S. J. Culver, Associate Keeper of the NHM Palaeontology Department, for their kind hospitality in their institution, and Dr A. R. Milner (Birkbeck College, University of London), who offered many helpful suggestions, gave much encouragement and revised a version of the manuscript. Thanks also go to Mr D. N. Lewis (NHM) who provided access to many facilities. Drs P. L. Forey and A. B. Smith (NHM) gave invaluable help and their suggestions improved the phylogenetic section of this paper. Dr P. E. J. Daley (formerly of the NHM) processed the phylogenetic data independently with the program Hennig86 and devoted much time to stimulating discussions. We are grateful to Mr J. J. Savill, who made his splendid collection available for study, Mr R. D. F. Oosthuizen, who found the holotype of Bokkeveldia oosthuizeni, the staff of the Palaeontology Laboratory and of the Photographic Unit of the NHM. Dr D. B. Norman (Sedgwick Museum, Cambridge) arranged for the loan of an important specimen. Mrs M. Joubert (South African Museum, Cape Town) lent additional material in her care. Special thanks go to Mr B. Lefebvre and Miss M. Marti Mus for their encouragement and lively discussions with MR and to Drs T. A. Elliott and P. D. Taylor (NHM) for their generous help and their patience in many circumstances. MR is grateful to the following people at the NHM Palaeontology Department for their encouragement during the months in which this work was carried out: Dr P. E. Ahlberg, Mrs D. Clements, Miss S. Evans, Dr R. A. Fortey, Miss T. Foster, Prof. B. G. Gardiner, Miss C. Jeffery, Dr T. Littlewood, Miss C. Mellish, Mr N. Monks, Dr N. J. Morris, Dr C. Patterson, Dr. J. Richardson, Mr A. Ross, Miss H. Santler, Mr C. Shute, Mr K. J. Tilbrook, Dr M. J. Weedon, Mr P. Whybrow and all those people whose names have been inadvertently omitted. This work forms part of a Ph.D. project on the evolutionary history of the mitrates carried out by MR at the University of London (Birkbeck College). MR gratefully acknowledges the receipt of a grant from the European Community (Training and Mobility of Researchers Programme). REFERENCES bain, a. G. 1856. Geology of South Africa. Eastern Province Monthly Magazine , 1, 396-407. barrande, j. 1887. Systeme silurien du centre de la Boheme, volume 7. Classe des Echinodermes. Ordre des Cystidees. Rivnac, Prague, 233 pp. bather, f. a. 1900. Chapters 8-12. 1-216. In lankester, e. r. (ed.). A treatise on zoology. Part III: Echinoderma. Oxford University Press, London, 344 pp. beisswenger, m. 1994. A calcichordate interpretation of the new mitrate Eumitrocy Stella savilli from the Ordovician of Morocco. Paldontologische Zeitschrift, 68, 443^162. boucot, a. J. 1971. Malvinokaffric Devonian marine community distribution and implications for Gondwana. Ana is da Academia Brasileira de Ciencias, 43 (Supplement), 23 49. caster, K. E. 1952. Concerning Enoploura of the Upper Ordovician and its relation to other carpoid Echinodermata. Bulletins of American Paleontology , 34, 1-47. — 1954. A Devonian placocystoid echinoderm from Parana, Brazil. Paleontologia do Parana (Centennial Volume), 137-148. — 1983. A new Silurian carpoid echinoderm from Tasmania and a revision of the Allanicytidiidae. Alcheringa, 7, 321-335. — and gill, e. d. 1967. Family Allanicytidiidae, new family. S561-S564. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part S. Echinodermata 1(2). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 352 pp. chauvel, j. 1941. Recherches sur les Cystoides et les Carpoides Armoricains. Memoires de la Societe Geologique et Mineralogique de Bretagne , 5, 1-286. — 1981. Etude critique de quelques echinodermes stylophores du Massif armoricain. Bulletin de !a Societe Geologique et Mineralogique de Bretagne , 13, 67-101. clarke, j. m. 1913. Fosseis devonianos do Parana. Monographias do Servico Geologico e Mineralogico do Brasil 1, 1-353. corstorphine, G. s. 1896. Report of the acting geologist for the year 1897. Annual Report of the Geological Commission of the Cape of Good Hope , 4-11. craske, a. j. and Jefferies, r. p. s. 1989. A new mitrate from the Upper Ordovician of Norway, and a new approach to subdividing a plesion. Palaeontology , 32, 69-99. cripps, a. p. 1989. A new genus of stem chordate (Cornuta) from the Lower and Middle Ordovician of Czechoslovakia and the origin of bilateral symmetry in the chordates. Geobios. 22, 215-245. — 1990. A new stem craniate from the Ordovician of Morocco and the search for the sister group of the craniata. Zoological Journal of the Linnean Society , 100, 27-71 240 PALAEONTOLOGY, VOLUME 40 1 99 1 . A cladistic analysis of the cornutes (stem-chordates). Zoological Journal of the Linnean Society , 102, 333-366. — and daley, p. e. j. 1994. Two cornutes from the Middle Ordovician (Llandeilo) of Normandy, France, and a reinterpretation of Milonicystis kerfornei. Palaeontographica, Abteilung A, 232, 99-132. dehm, r. 1932. Cystoideen aus dem rheinischen Unterdevons. Neues Jahrbuch fur M inera logic. Geologic und Paldontologie , Bei/Bd , Abteilung A, 69, 63-93. gill, e. d. and caster, k. e. 1960. Carpoid echmoderms from the Silurian and Devonian of Australia. Bulletins of American Paleontology, 41, 5-71. grobben, K. 1908. Die systematische Einteilung des Tierreiches. Verhandlungen der Zoologisch-Botanischen Gesellschaft in Wien , 58, 491-51 1. haeckel, O. 1 896. Die Amphorideen und Cystoideen. Beitrage zur Morphologie und Phylogenie der Echinodermes. Festschrift zur Siebzigsten Geburtstage von Carl Gegenbaur, volume 1. Engelmann, Leipzig, 179 pp. haude, r. 1995. Echinodermen aus dem Unter-Devon der argentinischen Prakordillere. Neues Jahrbuch fur Geologie und Paldontologie , Abhandlungen , 197, 37-86. miller, n. and theron, j. n. 1988. Benthic communities in the South African Devonian. 229-242. In McMillan, n. J., embrey, A. F. and glass, D. j. (eds). Devonian of the world , volume 3: palaeontology, palaeoecology and biostratigraphy. Memoir 14. Canadian Society of Petroleum Geologists, 714 pp. hillis, D. M. and bull, j. j. 1993. An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Systematic Biology , 42, 182-192. jaekel, o. 1901. Uber Carpoideen, eine neue Klasse von Pelmatozoen. Zeitschrift der Deutschen Geologischen Gesellschaft, 52, 661-677. jefferies, R. p. s. 1967. Some fossil chordates with echinoderm affinities. 163-208. In millot, n. (ed.). Echinoderm biology. Academic Press, London, 240 pp. — 1968. The Subphylum Calcichordata (Jefferies 1967) - primitive fossil chordates with echinoderm affinities. Bulletin of the British Museum ( Natural History), Geology Series, 16, 243-339. 1973. The Ordovician fossil Lagynocystis pyramidalis (Barrande) and the ancestry of amphioxus. Philosophical Transactions of the Royal Society of London, Series B, 265, 409 469. — 1979. The origin of chordates - a methodological essay. 443 — 477. In house, m. r. (ed.). The origin of major invertebrate groups. Systematics Association Special Volume 12. Academic Press, London, 515 pp. 1981a. In defence of the calcichordates. Zoological Journal of the Linnean Society, 73, 351-396. 1981A Fossil evidence on the origin of the chordates and echinoderms. 487-561. In ranzi, s. (ed.). Origine dei grandi phyla dei metazoi. Atti dei Convegni Lincei, 49, 1-565. — 1984. Locomotion, shape, ornament and external ontogeny in some mitrate calcichordates. Journal of Vertebrate Paleontology, 4, 292-319. — 1986. The ancestry of the vertebrates. British Museum (Natural History), London, 376 pp. — 1990. The solute Dendrocystoides scoticus from the Upper Ordovician of Scotland and the ancestry of chordates and echinoderms. Palaeontology, 33, 631-679. — 1991. Two types of bilateral symmetry in the Metazoa: chordate and bilaterian. 94-127. In bock, g. r. and marsh, j. (eds). Biological asymmetry and handedness. John Wiley and Sons, Chichester, 327 pp. — and lewis, d. n. 1978. The English Silurian fossil Placocystites forbesianus and the ancestry of the vertebrates. Philosophical Transactions of the Royal Society of London, Series B , 282, 205-323. — and prokop, r. j. 1972. A new calcichordate from the Ordovician of Bohemia and its anatomy, adaptations and relationships. Biological Journal of the Linnean Society, 4, 69-115. jollie, m. 1982. What are the ‘Calcichordata’? and the larger question of the origin of chordates. Zoological Journal of the Linnean Society, 75, 167-188. kolata, D. r. and jollie, M. 1982. Anomalocystitid mitrates (Stylophora, Echinodermata) from the Champlainian (Middle Ordovician) Guttenberg Formation of the Upper Mississippi Valley Region. Journal of Paleontology, 56, 531-565. — frest, t. j. and mapes, R. h. 1991. The youngest carpoid: occurrence, affinities and life mode of a Pennsylvanian (Morrowan) mitrate from Oklahoma. Journal of Paleontology , 65, 844—855. koninck, M. l. de 1869. Sur quelques echinodermes remarquables des terrains paleozoiques. Bulletin de T Academic Royale des Sciences Belgique, 28, 544—552. oosthuizen, R. d. f. 1984. Preliminary catalogue and report on the biostratigraphy and palaeogeographic distribution of the Bokkeveld fauna. Transactions of the Geological Society of South Africa , 87, 125-140. parsley, r. l. 1991. Review of selected North American mitrate stylophorans (Homalozoa: Echinodermata). Bulletins of American Paleontology, 100, 5-57. peterson, K. j. 1994. The origin and early evolution of the Craniata. 14-37. In prothero, d. r. and schoch. RUTA AND THERON: DEVONIAN MITRATES 24 i R. M. (eds). Major features of vertebrate evolution. Short courses in paleontology 7. The Paleontological Society and the University of Knoxville, Tennessee, Knoxville, 270 pp. 1995. A phylogenetic test of the calcichordate scenario. Lethaia , 28, 25-38. philip, G. m. 1979. Carpoids - echinoderms or chordates? Biological Reviews , 54, 439 — 47 1 . - 1981. Notocarpos garratti , gen. et sp. nov., a new Silurian mitrate carpoid from Victoria. Alcheringa , 5, 29-38. reed, f. r. c. 1925. Revision of the fauna of the Bokkeveld beds. Annals of the South African Museum , 22, 27-226. rennie, J. v. l. 1936. On Placocystella , a new genus of cystids from the Lower Devonian of South Africa. Annals of the South African Museum , 31, 269-275. salter, J. 1856. Description of Palaeozoic Crustacea and Radiata from South Africa. Transactions of the Geological Society of London , 7, 215-224. schwarz, E. h. l. 1906. South African Palaeozoic fossils. Records of the Albany Museum , 1, 347-404. swofford, D. L. 1993. PAUP : Phylogenetic Analysis Using Parsimony. Version 3.1.1 . Illinois Natural History Survey, Champaign, Illinois, 257 pp. tankard, a. J. and barwis, J. h. 1982. Wave-dominated deltaic sedimentation in the Devonian Bokkeveld basin of South Africa. Journal of Sedimentary Petrology , 52, 959-97 4. theron, j. n. 1970. A stratigraphical study of the Bokkeveld Group (Series). 197-204. In haughton, e. h. (ed.). Second International Gondwana Symposium , Proceedings and Papers. Council for Scientific and Industrial Research, Pretoria, 689 pp. 1972. The stratigraphy and sedimentation of the Bokkeveld Group. Unpublished D.Sc. thesis. University of Stellenbosch, 175 pp. - and Johnson, m. r. 1991. Bokkeveld Group (including the Ceres, Bidouw and Traka Subgroups). 3-5. In JOHNSON, M. R. (ed.). SACS Catalogue of South African Lithostratigraphic Units , 3, 52 pp. and loock, J. c. 1988. Devonian deltas of the Cape Supergroup, South Africa. 729-740. In McMillan, n. J., embry, A. F. and glass, D. J. (eds). Devonian of the world , volume 1 : regional synthesis. Memoir 14, Canadian Society of Petroleum Geologists, 795 pp. and thamm, a. G. 1990. Stratigraphy and sedimentology of the Cape Supergroup in the Western Cape. Guidebook Geocongress '90. Geological Society of South Africa, 64 pp. thom, G. 1830. Remarks on the geology of South Africa. South African Quarterly Journal , 1, 269-27 1. UBAGHS, G. 1967. Stylophora. S496-S565. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part S. Echinodermata 1(2). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 352 pp. - 1971 . Diversite et specialisation des plus anciens echinoderm.es que Ton connaisse. Biological Reviews , 46, 157-200. 1975. Early Palaeozoic echinoderms. Annual Review of Earth and Planetary Sciences , 3, 79-98. 1981. Reflexions sur la nature et la fonction de l’appendice articule des carpoides Stylophora (Echinodermata). Annales de Pa/eontologie. Invertebres , 67, 33^48. wetherby, a. G. 1879. Description of a new family and genus of lower Silurian Crustacea. Journal of the Cincinnati Society of Natural History , 1, 162-166. M. RUTA Department of Biology Birkbeck College Malet Street, London WC1E 7HX, UK and Department of Palaeontology The Natural History Museum Cromwell Road, London SW7 5BD, UK J. N. THERON Typescript received 29 November 1995 Revised typescript received 7 May 1996 Geological Survey P.O. Box 572 Bellville, Cape Town South Africa 242 PALAEONTOLOGY, VOLUME 40 APPENDIX 1 List of the characters used in the phylogenetic analysis (discussion in the text). The character-states are indicated by numerals in brackets. The plesiomorphic condition is indicated by 0, whereas the derived states are indicated by I or 2. 1. Dorsal fore-tail plates in contact with ventral fore-tail plates along flexible (0) or rigid (1) sutures. 2. Presence (1) or absence (0) of thickening along the posterior margin of the dorsal fore-tail plates. 3. Presence (1) or absence (0) of laterally expanded styloid blades. 4. Presence (0) or absence (1) of dorsal longitudinal canal in the hind-tail ossicles. 5. Ventral hind-tail plates forming a keel (0) or a blunt surface (1). 6. Plates e and 9 contact laterally small, scale-like ventral plates comprised between them and the rearmost lateral marginal elements (0) or the rearmost lateral marginal plates only (1). 7. Posterior, ventral tessellated bar present with the median elements separated by the interposition of plate p (0), present but with plate p posterior to the median elements (1) or absent (2). 8. Ventral head skeleton made up partly (0) or entirely (1) of tessellated plates. 9. Ventral head plates partly (0) or entirely (1) arranged in rows. 10. Plates £ and 9 completely separated by p (1) or not (0). 11. Presence (1) or absence (0) of VI 7. 12. Second ventral row, if present, made up of lateral elements only. 13. Third ventral row, if present, made up of lateral elements only. 14. Fourth ventral row, if present, made up of lateral elements only. 15. Presence (1) or absence (0) of first ventral row. 16. Plate V12 absent (0) or present and not expanded posteriorly up to the level of the third ventral row (1) or expanded (2) and in contact with plate p (3). 17. Plates £ and 0 wider than long (0) or not (1). 18. Postero-median margin of s and 9 straight (0) or curved (1). 19. Presence (0) or absence (1) of lateral line opening. 20. Ventral ribs absent (0) or present and not extending (1) or extending (2) in front of plates £ and 9. 21 . Ventral riblets absent (0) or confined to the fourth ventral row only (1) or present also in front of the latter (2). 22. Presence (1) or absence (0) of oral spines. 23. Presence of a thickening along the foremost margin of the lateral plates of the second ventral row. 24. Presence of three (0) or two (1) centro-dorsal plates. 25. Head outline asymmetrical (0) or symmetrical (1). 26. Mouth opening facing leftward (0) or perpendicular to the longitudinal axis of the body but not bisymmetrically divided by it (1) or divided by the axis (2). 27. Presence of six (0) or five (1) left marginal plates. 28. Presence of six (0) or five (1) right marginal plates. 29. Plate n much larger than plates b and c. 30. Anterior region of dorsal skeleton flexibly articulated with anterior dorsal marginal and centrodorsal plates. 31. Presence (0) or absence (1) of oral platelets. 32. Plates b and c polygonal (0) or wedge-shaped and with major axis oriented obliquely (1). 33. D10 separates D12 from the left series of marginal plates. 34. D10 and D12 roughly equal in width (1) or not (0). 35. Posterior projections of h and i visible in ventral view (0) or not (1). 36. Presence (1) or absence (0) of dorsal ribs. 37. Presence (1) or absence (0) of dorsal riblets. 38. Presence (1) or absence (0) of extensive resorption fields over the left and the right pharynx. APPENDIX 2 Data matrix. Symbols as follows: ?, missing information; n, not applicable character; u, uncertain state assignment. RUTA AND THERON: DEVONIAN MITRATES 243 Character number Taxa 1 1 1 1 1 1 I 1 1 1 1234567890123456789 Allanicytidium flemingi A teleocystites guttenbergensis Australocystis langei Barrandeocarpus jaekeli Barrandeocarpus norvegicus Eumitrocystella savilli Mitrocystella barrandei Mitrocystella incipiens Notocarpos garratti Pkicocystella africana Placocystites forbesianus Tasmanicytidium burretti Victoriacystis wilkinsi ?11??12111011103111 1101111110100011111 ?????? ???■?? 9 ?? n ii ? n 1107111000100000101 1101111100100010111 70001 10000000000000 0000007000000000000 00000000000000001 10 1117112111011103111 1111112111011103111 1101111110101012111 77777121 1 101 1 1031 1 1 1107111110110012111 Character number Taxa 2222222222333333333 0123456789012345678 Allanicytidium flemingi Ateleocystites guttenbergensis Australocystis langei Barrandeocarpus jaekeli Barrandeocarpus norvegicus Eumitrocystella savilli Mitrocystella barrandei Mitrocystella incipiens Notocarpos garratti Placocystella africana Placocystites forbesianus Tasmanicytidium burretti Victoriacystis wilkinsi 7 7 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 7 2010012110000001101 7 7 1 1 1 1 2 1 1 1 1 1 1 1 1 1 7 7 7 2100001 1 10000001 10? 20000121 10000001 101 OOOOlOOlOnOOOOOlOOl 0000000000000001000 1000000100000000000 201111211111100110? 2211112111111101111 20101 121 10000001 101 221111211111110100? 20101 121 1017000177? PERMIAN S UP A I A FRONDS AND AN ASSOCIATED AUTUNIA FRUCTIFICATION FROM SHANXI, CHINA by WANG ZI-QIANG Abstract. Three new species of bipartite Supaia frond, one associated with an Autunia ovuliferous organ, have been identified. S. shanxiensis sp. nov. from Central Shanxi is roughly similar in gross morphology to Supaia species from the Permian Hermit Shale of North America, but differs in the size, shape and other features of the frond. It can also be distinguished by the pinnules having an entire margin, a faintly decurrent base and being more closely spaced, and the venation consisting of a weak mid-vein. S. contracta sp. nov., from North Shanxi, is characterized by pinnules with a markedly constricted base and distinctive cuticles, and compares closely with some of the Upper Permian Tatarina species from the Urals. The Southern Shanxi species, S. yuanquensis sp. nov., is distinguishable from both the above in its smaller frond with a delicate primary rachis, its strongly decurrent and elongate pinnules, and its thin lamina texture. Autunia shanxiensis sp. nov. shows individual cones consisting of bilaterally symmetrical megasporaphylls in spiral attachment, in contrast to the lax, modified, fertile pinnae of A. milleryensis. Although not found in attachment, the close association of S. shanxiensis fronds and A. shanxiensis ovuliferous organs suggests that, at least in North China, the Supaia- type frond may have belonged to the Peltaspermaceae. It is argued that the names Autunia and Peltaspermum, originally proposed as ‘organ’-genera, should not be used for natural genera based on partly reconstructed fossil plants. Abundant fungal-spots on or within Supaia fronds in North China are evidence of the increasing deterioration of the environment during the Permian. Supaia White, 1929 has been regarded as endemic to the Permian Hermit Shale of the south- western United States (Read and Mamay 1964; Chaloner and Meyen 1973; Meyen 1987; Lemoigne 1988; Cleal and Thomas 1991; DiMichele and Hook 1992). However, there is some evidence that it also occurs elsewhere in the Northern Hemisphere, such as Lodeve in southern France (Doubinger and Heyler 1975), Moravia in the Czech Republic (Nemejc and Augusta 1937; Dijkstra 1965), the Kuznetsk Basin of Siberia (Zalessky and Tchirkova 1935; Neuburg 1948), and the northern periphery of China (Huang 1977; Zhou and Zhou 1986). Significantly, all previously reported Chinese specimens came from north of the Tianshan-Xinggan Suture, and are thus from the Angaran terrane. Hitherto, our knowledge of the genus has been restricted to its gross morphology, with nothing known of the fructification or cuticles. Thus, its affinities have been uncertain. Between 1987 and 1991, when studying the phytostratigraphy of the Permian red-beds in North China, 1 recognized that most specimens previously assigned to Protoblechnum or Compsopteris from the Tianlongsi Formation had bipartite fronds with two monopinnate branches, like Supaia. Later professional collections made at a locality in the Qingyuan District of Shanxi (Wang-tao village) eventually yielded many entire bipartite Supaia- type fronds, associated with Autunia- type ovuliferous organs. This previously unknown obligate association suggests that both belong to one plant. Fronds of another type of Supaia from Xuangan Coal-Mine have yielded well-preserved cuticles and particularly exhibit many fungal-spots. Although some of these specimens have been illustrated previously (Wang 1996) the present paper provides the first full documentation. | Palaeontology, Vol. 40, Part 1, 1997, pp. 245-277, 6 pls| © The Palaeontological Association 246 PALAEONTOLOGY, VOLUME 40 text-fig. 1 . Localities in Shanxi yielding Supaia from the upper Tianlongsi Formation. STRATIGRAPHICAL SETTING The plant-bearing strata belong to what until recently was known as the ‘Upper Shihhotze Formation', but which was renamed the Tianlongsi Formation to avoid a nomenclatural duplication with the Shihhotze Group. This lithostratigraphical unit was re-defined by the 212 Geological Survey Team of Shanxi Province (Chen and Niu 1993). I have recognized three plant- assemblage zones within the Tianlongsi Formation in its type section in the Western Flill of Taiyuan, Shanxi in ascending order: the Plagiozamites Zone, the Chiropteris Zone (or subzone) and the Psygmophyllum Zone. In the north of Taiyuan, the Chiropteris Zone is absent. Of these zones, the Psygmophyllum Zone has hitherto not been studied in detail, although Halle (in Norin 1924, p. 23) had referred to it as ‘the uppermost plant-bearing zone of the Upper Shihhotze Series with Psygmophyllum and Taeniopteris'. A summary of the Psygmophyllum Zone floras is given in Appendix 1. All plant-bearing biostromes in the Psygmophyllum Zone are small lenses of grey-greenish muddy or silty sandstone about 10-50 mm thick and 1-2 m in lateral extent. The lenses are within or under the two to three sets of white-greyish channel sandstones or sandy conglomerates (1-20 m thick), which in turn form large intercalations within thick red mudstones and represent ephemeral channel deposits in a seasonally alternating arid-wet climate. The fossils only rarely represent autochthonous or parautochthonous burials, in which the plants can easily be identified at the species level. Most are allochthonous or even fragmentary burials, where the plants can be barely assigned at the generic level based on their gross morphology. In all the sections (Text-fig. 1), the Supaia-bzaring biostromes are restricted to the uppermost part of the Tianlongsi Formation. Among them, only those at the Qinyuan ( = Wangtao), Xuangan and Yuanqu sections are parautochthonous burials, the rest being fragmentary ones. In general, the total plants in each biostrome are of rather low diversity, not exceeding ten species belonging to five to seven genera. Wangtao section (Text-fig. 2). The Supaia- bearing lens is a thin grey mudstone or siltstone, lying below a 2m thick white-yellowish sandstone, and is about 40 m from the top of the formation WANG: PERMIAN SUPAIA AND AUTUNIA 247 — 300m G 03 • H a cd n cd W A fv Psygmophyllum Zone zzz^ -200 ^ Chiropteris reniformis & fs J Subzone 3 tfl be G 'g ctf 100 A f3 k U > Plagiozamites Zone + + + + + + + + + ++ + + + + + + + + + + + + + + + + + + + 20m r Z\fe Plant-assemblage Zones plant-bearing bed No. mudstone .'.*.1 sandstone conglomerate 5 1 § S -9 4* 3 .0] •fi 5 1 = J1 ” ~ o: ^ ^ k) ^ ■9 C JP -S S a '3 3 C! ^ S .o S 5j •§ ■» & § ^ l &■$. 9 3 • 2 2^-2 Q) <3 o 53 ^ cj r-i *3 s ^ CJ S S 2. .®s s s § § i 2 .b1 9 •3 b | ■§, g. | P S * S; 8 £ text-fig. 2. Distribution of main plant fossil form-genera in the Tianlongsi Formation at the Wangtao section, Qinyuan, Shanxi. text-fig. 2. Distribution of main plant fossil form-genera in the Tianlongsi Formation at the Wangtao section, Qinyuan, Shanxi. Asselian Tianlongsi Formation lyr> Lepidoden dron + + Ann ularia-Sphenophyllum 4 + + Loba tann ularia + 4- + Plagiozamites + + Pecopteris-Sph en opteris + + 4- 4- + Fascipteris 4- Odontopteris + Protoblechn urn 4- + Gigantonoclea 4- Taenioptens + 4 4 4 Lesleya + + Supaia-Autunia Cb iropteris-Nystroem ia 4 4 Psygm ophyllum 4 WANG: PERMIAN SUPAIA AND AUTUNIA 248 PALAEONTOLOGY, VOLUME 40 (Text-fig. 2, plant-bearing bed No. 7). The lens comprises two biostromes: the upper one with abundant Psygmophyllum multipartitum Halle and occasional Sphenopteris cf. gothanii Halle; the lower one containing rich Supaia , and rare ‘ Chiropteris kawasakii' Kon’no, Rhipidopsis sp., Nystroemia pectiniformis Halle, Lobatannularia heianensis (Kodaira) Kawasaki, Annularia shirakii Kawasaki, Lesleya sp .{= Taeniopteris cf. schenki Sterzel). Supaia shanxiensis sp. nov. and associated Autunia shanxiensis sp. nov. are well preserved but unfortunately lack cuticle. Xuangan section. The biostrome here is a thin, monocyclic, silty shale, 01-0-2 m thick. It occurs in the second from top set of grey-greenish sandstone, which is about 15 m thick, and is about 80 m from the top of the formation. The fossils are an parautochthonous burial, representing plant foliage that was transported for short distance. The main identifiable plants are Supaia contracta sp. nov., Psygmophyllum multipartitum Halle , Fascipteridium ellipticum Zhang and Mo, Lobatannularia multifolia Kon’no and Asama, Pecopteris cf. feminaeformis (Schlotheim) Sterzel, P. arcuata Halle, Taeniopteris densissima Halle, T. szeiana Chow, Lesleya sp. and Peltaspermum sp. Among these, Supaia contracta is significant in having well-preserved cuticles, which hitherto have not been known from this genus. Yuanqu section. The Supaia- bearing biostrome here is in the middle-upper part of the formation, probably at a lower level than those in the previous sections. The main plant fossils are parautochthonously buried pteridosperms, including Psygmophyllum multipartitum Halle, Chiropteris reniformis Kawasaki, Callipteris changii Sze, Supaia yuanquensis sp. nov., Neuropteridium coreanicum Koiwai and Gigantonoclea sp. There are also rare allochthonous Lobatannularia multifolia Kon’no and Taeniopteris densissima Halle, amongst others. Institutional abbreviation. TIGM, Tianjin Institute of Geology and Mineral Resources. SYSTEMATIC PALAEONTOLOGY Order peltaspermales Nemejc, 1968 Satellite form-genus supaia White, 1929 Type species. Supaia thinnfeldioides White, 1929, p. 62. Diagnosis. Bipartite frond with two branches or pinnae, each of which is asymmetrical, with larger outer pinnules and smaller inner ones. Primary rachis forked near the stout base, producing two rigid, long branches. Pinnules vary from alethopteroid-like to neuropteroid-like in shape, either broadly attached to the rachis with decurrent base, or fairly constricted at the base prior to explanation of plate 1 Figs 1-8. Supaia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian; x 1. 1, 9306-1 (holotype); bipartite frond with medianly forked primary rachis. 2, 6, partial frond with normal pinnules attaching to a slender rachis; 2, 9306-8; 6, 9306-4. 3, 9306-16; part of forked primary rachis covered with longitudinally vascular striation and with several remnants of pinnules attached. 4, 9306-2; small whole frond. 5, 9306-7; small frond with thickened base of the primary rachis to which a few reduced pinnules are attached. 7, 9306-9; part of bipartite frond, showing the thickened base of its primary rachis. 8, 9306-12; smaller frond, with a well- developed pinna on the right and a reduced one on the left, and the shortened base of its primary rachis. PLATE 1 WANG, Supaia 250 PALAEONTOLOGY, VOLUME 40 text-fig. 3. The lower parts of various Supaia shanxiensis sp. nov. fronds; Tianjin Institute of Geology and Mineral Resourses. a, 9306-2; b, 9306-12; c, 9306-7; D, 9306-9; E, 9306-1 (holotype); f, 9306-4; G, 9306-3; H, 9306-31; i, 9306-5; J, 9306-16. All x 0 92 WANG: PERMIAN SUPAIA AND AUTUNIA 251 text-fig. 4. Variation in ranges of pinnule size between: a, North American Supaia- type fronds; b, Supaia shanxiensis sp. nov. fronds; c, North American Brongniartites fronds. Sizes of the American fronds are estimated from the photographs illustrated in White (1929). C*J North American Supaia -type fronds, (X) Supaia shanxiensis fronds, °. North American Brongniartites fronds. attachment. The near-terminal pinnules are connate, lobed or pinnatifid. Below the main fork of the frond are three to five pairs of more or less reduced, triangular or semicircular pinnules. Laminae thick, coriaceous in texture. Midvein well-developed. Lateral veins generally extend from midvein at 45°, fork one or twice, and are often hidden in the pinnule lamina. Supaia shanxiensis sp. nov. Plates 1-3; Plate 6, figures 6-9; Text-figures 3, 5 1989 Protoblechnum wongii Halle; Si ( non Halle), p. 56, pi. 65, fig. 2. 1996 Supaia sp. a Wang, pi. 1, figs 1-2, 6; pi. 2, figs 3, 5. Derivation of name. From Shanxi province, where the main locality for the plant occurs. Holotype. TIGM 9306-1 (PI. 1, fig. 1); upper Tianlongsi Formation, Wangtao village, Qinyuan, Shanxi. Diagnosis. Frond of medium size, about 100 mm long and 70 mm wide, bipartite with two equal or unequal branches, each one lanceolate-oblong, asymmetrical, with its outer side wider than the inner one. Primary rachis forked into two rigid branchs at acute (20^40°) angle. A few pairs of reduced or undeveloped pinnules attached to both sides of the basal part of frond, below the forked 252 PALAEONTOLOGY, VOLUME 40 point. Pinnules alternate or subopposite, closely spaced. Pinnules callipteroid-alethopteroid in shape, about 5-75 x 6-26 mm in size, with obtuse apex and slightly decurrent base, and entire or sometimes undulate margins. Outer pinnules of a branch markedly larger than inner ones, especially near the fork of the primary rachis. Pinnules near terminal connate, forming a rather large, obtuse apex. Midvein well-developed, nearly reaching to the apex in larger pinnules but faint in smaller ones. Lateral veins generally obscure, concealed within the thick lamina; they extend obliquely from the midvein at about 45° and then bifurcate once or twice. Several veins extend immediately from the rachis. Description (gross morphology). The whole frond has an obovate or elongatedly triangular form, varying from 80 mm long by 40 mm wide, to 150 mm long by 100 mm wide. The primary rachis is forked, with two equal or unequal branches arching away from one another at 15-40°. Below the fork, the primary rachis is 10-70 mm long and 2-7 mm thick, strong and woody, covered with fine striations, and slightly widens towards its proximal end. The fronds illustrated on Text-figure 3e-g have two roughly equal branches, but those on Text- figure 3a-c clearly show two unequal branches, especially that shown on Text-figure 3b, which has a normal pinna on the right and an undeveloped (or abortive) pinna on the left. As shown in Text-figure 3, most fronds are not strictly speaking bisymmetrical. The branches (or pinnae) are monopinnate, lanceolate or oblong-oboval, usually 60-100 mm long and 15-65 mm wide, but with a maximum length of over 120 mm and width of 65 mm. The branches are asymmetrical, with pinnules larger on the outer side than on the inner side. The broadest portion of the branch is in its upper part. The pinnules vary from 5 mm long by 6 mm wide to 75 mm long by 26 mm wide (Text-fig. 4). They are mostly of an alethopteroid-type, oblong or obtusely triangular, with a rounded apex and fully expanded or decurrent base; only rarely are slightly constricted pinnules present. Nine to 12 pairs of pinnules occur on each branch, alternately or suboppositely arranged, closely spaced or a little overlapping, and vertically or obliquely attached to the rachis. Pinnule margins are entire or sometimes slightly undulate. Towards the distal end of the branches, adjacent pinnules gradually become connected to each other to form an oboval or broad fan-like terminal pinnule (PI. 2, fig. 1). Near the base of the frond, three to five pairs of pinnules are attached to the primary rachis, sparsely spaced, and more or less reduced or scarious. The pinnule lamina is rather thick and rugose; it appears uneven or may have been scabrous prior to being buried. Table 1 shows the main measurements from some selected specimens of the new species. Venation is alethopteroid or callipteroid, with a moderately thick but (on the surface) faint midvein, which dissolves about two-thirds of the way along the pinnule. Ten to 15 pairs of lateral veins are produced obliquely by the midvein at an acute angle, and are always concealed within the lamina. Some veins in the basal portion of a pinnule arise immediately from the rachis. Many striped rusts occur along the lateral veins (PI. 3, figs 5-6) as result of fungal infection. Description (microscopic features). Unfortunately, virtually no cuticles can be extracted from the type specimens from Wangtao section, to show the epidermal characters of the new species. However, some minute fragments of residue left on the specimens can provide some limited information. For example, a fragment of cuticle illustrated on Plate 6, figure 6 shows the skeleton of an epidermal cell with spines extending from the EXPLANATION OF PLATE 2 Figs 1-7. Supaia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian. 1, 9306-24; apex of pinna showing widening terminal pinnule; x 1. 2-5, variation in pinnule morphology; x 2. 2, 9306-28; normal pinnule showing clear venation and attachment to rachis along its entire decurrent base. 3, 9306-30; two pinnules attached vertically to the rachis, showing many splits along the veins, which had developed prior to burial 4, 9306-45; semi-round pinnule attached to the basal part of primary rachis. 5, 9306-27; lanceolate pinnule attached to rachis with slightly constricted base. 6, 9306-60; part of large pinna; x 1. 7, 9306-3; smaller bipartite frond, with asymmetrical branches; x 1. PLATE 2 WANG, Supaia 254 PALAEONTOLOGY, VOLUME 40 table 1. Dimensions of main specimens used in the description of Supaia shanxiensis sp. nov. Primary rachis Length x width (mm) Specimen Size of frond length x width Angle of primary of branches Size of largest number (mm) (mm) rachis dichotomy produced by dichotomy pinnules (mm) 9306-1 (a) 110 X 100 60 X 7 35° 60 X 30 30 X 12 -1 (b) 100 X 90 40 X 5 30° 60 X 40 12 X 15 _2 80 X 40 20 X 3 20° 60 X 19 12 X 6 -3 100 X 80 25 X 7 40° 80 X 45 25 X 15 -4 120 X 60 75 X 5 15° - 30 X 15 -5 - 20 X 6 20° - 22 X 13 -6 80 X 90 - 35° 100 X 60 35 X 13 -7 - - 30° - 5 X 6 -8 100 X 70 20 X 2 40° 80 X 40 20 X 10 -9 - 40 X 6 25° - 20 X 12 -10 - 10 X 6 - - 5 X 8 -12 70 X 60 - - 50 X 27 12 X 6 -17 - - O O 'T 100 X 80 50 X 18 -22 - - - 110 X 60 34 X 16 -25 - - - 50 X 30 55 X 19 -26 - - - - 75 X 26 -45 - 20 X 2 o O - 17 X 20 corner of the cell wall in the inner view. More importantly, the SEM photograph on Plate 6, figure 5 shows clearly one of the small, dark spots along both sides of veins that represent fungal bodies. These spots are also visible in most hand specimens (PI. 3, fig. 5). Comparison with North American species. There is no fundamental difference at the generic level between the present specimens and material from the Hermit Shale in America. They both have (1) a distinctive bipartite frond architecture with two monopinnate branches; (2) alethopteroid-type pinnules with a thick coriaceous texture; (3) pinnules of similar appearance on both sides of the branches and below the fork of the primary rachis; and (4) oblique lateral veins concealed within the laminae. Within this genus, White (1929, pp. 54-86) described 12 species (three of which are indeterminable) from the Hermit Shale in Grand Canyon, Colorado. He also attributed some other bipartite fronds from there to Brongniartites Zalessky, including B.l aliena White (pi. 27, fig. 2) and B. ? yakiensis White (pi. 28, fig. 3). In my opinion, all of these species may belong to one genus and even one species. Most are not strictly delimited on the basis of a constant feature and those which are separated on more clear-cut features (e.g. size, shape, base and apex of a pinnule) are rare (e.g. Supaia rigida White, S. anomala White, S. linearifolia White). EXPLANATION OF PLATE 3 Figs 1-6. Supaia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian. 1, 9306-61 ; part of forked-primary rachis from which pinnules had been shed, showing longitudinal vascular striation; x2. 2, 9306-47; isolated pinnule with fine veins, showing extensive base; x 2. 3, 9306-29; similar pinnule but with slightly contracted base; x 2. 4, 9306-5; lower part of bipartite frond, showing the thickened base of primary rachis below the fork; x 1. 5-6, parts of pinnae infected by fungi, showing many small fungal spots along both sides of lateral veins; x2; 5, 9306-44; 6, 9306-22. PLATE 3 WANG, Supcda 256 PALAEONTOLOGY, VOLUME 40 Among White’s species, only Supaia thinnfeldioides (White 1929, pi. 14) shows a close comparison with S. shanxiensis , having similarly closely spaced pinnules with a coriaceous lamina concealing the venation, and a similar overall architecture of the frond. However, the former differs in having much longer pinnules with a stronger midvein that reaches to the pinnule apex, and a marked expansion of the base of the pinnule with a decurrent distal margin. S. shanxiensis differs consistently from most of the other of White’s species. For example, the Chinese species have a much lower pinnule length :width ratio than the American ones (Text-fig. 4). The Chinese species also has a much less variable pinnule shape than the American ones, which vary from slender-linear to lanceolate. A few American fronds have markedly lobed pinnules, in contrast with the Chinese frond which has consistently entire margined (though sometimes a little uneven) pinnules. The base of the pinnules in these American species also tends to show more variation, from broadly decurrent to constricted, while the pinnules of S. shanxiensis are consistently fused to the rachis along their entire base, except below the fork of the main rachis where they may have a reduced or slightly constricted (on acroscopic side) base. The Chinese species consistently presents closely spaced pinnules, while a marked separation tends to occur between adjacent pinnules in the American fronds, although a few more closely spaced pinnules may also occur in the latter; the latter also clearly show an upper ‘puckered corner’ at the base of the pinnule, resulting from the base buckling or twisting prior to attachment. Finally, the midvein of the Chinese species is weakly developed and clearly differs from the strong, prominent midvein reaching the apex of the pinnule in the American species. Comparison with Asian species. A few records of Supaia- type fronds occur in the literature on the Permian floras of East Asia, especially from the Angaran Kingdom. In North China, several specimens assigned to Protoblechnum wongii Halle from the Tianlongsi Formation show the same type of distinctive bipartite frond with two branches bearing alethopteroid pinnules and a Supaia- like venation (Chow et al. 1955, pi. 1 ; Si 1989, pi. 65, fig. 2). Tchirkova (in Zalessky and Tchirkova 1935, p. 1108, fig. 6) described a new species, Supaia tomiensis Tchirkova, from the Upper Permian of the Kuznetsk Basin, Siberia (see also Neuburg 1948, pi. 42, fig. 4). It was based only on a small bipartite frond with a slender primary rachis but lacking the part below the fork. Compared with the S. shanxiensis , its pinnules are much smaller and seemingly, have a thinner lamina with a less developed midvein. Huang (1977) described two species of Supaia frond from the so-called ‘Permian Angaran floras’ in the Xiao Hinggan Range of North-east China. S. shenshuensis Huang shows a large frond but does not appear to have any pinnules below the fork of the main rachis. Its terminal pinnules are elongate, and more or less dichotomously formed, clearly differing from those of 5. shanxiensis. The other species, S. tieliensis Huang, has a frond more similar to S. shanxiensis, having closely spaced pinnules with an obtuse apex (Huang 1977, p. 46, pi. 36, figs 1-3; pi. 37, figs 1-2; pi. 38, figs 1-5; text-fig. 14). However, it differs from the latter in each branch bearing fewer pinnules (six to seven pairs), which are more oblique (50-60°) to the slender rachis and which have much denser lateral veins (9-10 veins per 10 mm). Also, its terminal pinnule is single, rather than connate as in other Supaia species. Zhou and Zhou (1986, p. 56, pi. 9, fig. 3) assign a specimen from the Upper Permian of Xingjian, China, to Supaia sp. In addition, the following specimens from the Kazanian of the Urals, probably belong to Supaia due to the bipartite architecture of the frond: Callipteris adzvensis Zalessky (Fefilova 1973, pi. 30, fig. 4), Callipteris bel/a Zalessky (Vladiminovich 1986, pi. 140, fig. 6), and Comia biarmica Zalessky (Vladiminovich 1986, pi. 149, fig. 4). However, these specimens are too fragmentary or poorly preserved to be identified accurately. WANG: PERMIAN SUPAIA AND AUTUNIA 257 text-fig. 5. Supaia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian. Parts of pinnae showing several blanks or gaps on the uneven surface, indicating that the lamina was desiccated prior to burial ; x 1 . a-b, upper parts of median pinnae; a, 9306-23; b, 9306-22; c-D, parts of large pinnae; c, 9306-26; D, 9306-25. Supaia contractu sp. nov. Plate 5; Plate 6, figures 1-5, 10; Text-figures 6-7 1996 Supaia sp. b Wang, pi. 2, fig. 5; pi. 3, fig. 9. Derivation of name. From the constricted nature of the pinnule base. Holotype. TIGM 8915-7; upper Tianlongsi Formation, Xuangan Coal Mine, Yuanping, Shanxi. The gross morphology of the main part of the bipartite frond is shown in Text-figure 6a, c; Plate 5, figures 1, 3-6 and Text-figure 7 show the cuticles. Diagnosis. Frond moderate in size, bipartite into two monopinnate asymmetrical branches, on which are developed strong outer pinnules and reduced inner ones. Primary rachis thick and strong. 258 PALAEONTOLOGY, VOLUME 40 and covered with dense woody striations. Pinnules neuropteroid-like, with a round apex and a considerably constricted base. Midvein well-developed and reaching to the apex of pinnule. Lateral veins fine, arching outwards, bifurcated once or twice, more or less fascicled. Cuticles amphistomatic, enclosing the thick hypodermis or mesophyll which tends to be infected by fungi bodies. Both epidermises are alike, but with minor differences in the shape of cell and stomatal index (i.e. about 8-9 in the upper cuticle and 11-12 in the lower one). Stomatal apparatus round, surrounded by a prominent rim formed by the elevated anticlinal wall of each subsidiary cell. Each stoma generally with more than five subsidiary cells, which are not so specialized that a distinctive area of subsidiary cells can not be recognized around each stoma. Guard cells sunken below a deep stomatal cavity. Papillae generally weakly developed, except locally. Description ( gross morphology). Of all the present specimens, only one (Text-fig. 6a) shows the main part of the bipartite frond. The primary rachis is thick and strong, and forked at about 40°, 3 mm wide above and 5 mm wide below the fork. This rachis is covered with longitudinal striations representing vascular strands, and some strumolose projections. Pinnules are general large, over 50 mm long and 15 mm wide, with an obtuse apex and a distinctively constricted base. Pinnule margins are entire and the lamina is thick, coriaceous but apparently easily infected by fungi. Venation is of a neuropteroid-type, but with the midvein dissolving near the apex; lateral veins are more or less fascicled, fine, costally arched, dense, and forked once or twice. Plate 6, figure 5 shows a piece of pinnule cuticle showing vein-like stripe-rusts, consisting of many small fungal spots along both sides of the lateral veins. The stripe-rust clearly illustrates the traces of the fascicled-style of venation, which is a characteristic of this new species. Description ( microscopic features). Cuticles are moderately thick, the upper one being slightly thicker than the lower one. The epidermis is amphistomatic, but the upper epidermis consists of more regularly rectangular cells with a lower stomatal index. The upper epidermis consists of isodiametric cells with even or smooth periclinal walls and poorly developed papillae, and does not show a marked differentiation between the costal and intercostal areas, nor between the ordinary and subsidiary cells. Stomata are random in arrangement, round to oblong-elliptical in view from the outer surface of the cuticle. In general, a stomatal apparatus is 20-30 //m in diameter. A highly elevated, thickened rim surrounds the aperture, formed by a raised part of the anticlinal wall of each subsidiary cell adjacent to the aperture. Guard cells are strongly cutinized, reniform especially along their dorsal portion, and sunken at the bottom of a deep cavity (Text-fig. 7g-i). Four to six subsidiary cells surround the guard cells, and are weakly distinguishable from the ordinary cells, so that a subsidiary cell ring or area is either invisible or only poorly visible (PI. 5, fig. 1 ; Text-fig. 7d). The stomatal index is normally constant at 8-10, but is sometimes much lower. The lower epidermis (PI. 5, fig. 3; PI. 6, fig. 1) differs mainly from the upper one in the ordinary cells being more or less elongate, in the occurrence of rare files of elongate cell along the veins (PI. 5, fig. 6) and in the occasional appearance of hollow papillae (PI. 6, fig. 1). The epidermis of the rachis (PI. 5, fig. 5) consists of elongate cell with rare stomata. Papillae are poorly developed. Fungal remains. Almost all the known fronds of this species were infected by fungi and bacteria prior to burial, and show fungal-stripe rusts. These are in the form of small strumolose dark-spots EXPLANATION OF PLATE 4 Figs 1-2, 4-9. Supaia yuanquensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Yaotou village, Yuanqu, Shanxi; upper-middle Tianlongsi Formation, Upper Permian. 1-2, bipartite fronds showing pinnules and forked primary rachis. 1 , 8806-1 (holotype); xl.2, 87Y5-1; x 1. 4, 7-9. various pinnules from adult fronds, showing obliquely extending lateral veins and strongly decurrent base; x 1. 4, 87Y5-19. 7, 87Y5-e. 8, 87Y5-9. 9, 87Y5-11. 5-6, 87Y5-S1; part of young bipartite frond showing rare oblique veins; 5, x 1; 6, x 2. Fig. 3. Supaia yuanquensis sp. nov.? Tianjin Institute of Geology and Mineral Resources; 87Y5-12; a questionable specimen, showing larger pinnules and denser lateral veins; Yaotou village, Yuanqu, Shanxi; upper-middle Tianlongsi Formation, Upper Permian; x 1. PLATE 4 WANG, Supaia 260 PALAEONTOLOGY, VOLUME 40 occurring along both sides of the veins (PI. 6, fig. 5). They are mostly multicellular, spherical, fungal bodies (PI. 5, fig. 2; Text-fig. 7b, right) but rarely are slender, monocellular mycelia adhering to a cell wall (Text-fig. 7a). A detailed description of these bodies will be published elsewhere. Comparison. Supaia contracta is characterized by pinnules with a markedly constricted base and a more or less fascicled venation. Significantly, this is the only species of Supaia known to date in which cuticular evidence is available. In the northern hemisphere, cuticular characters are known for many Permian peltaspermous genera, e.g. Ca/lipteris Brongniart non Bory (Barthel 1962; Barthel and Haubold 1980; Wang and Wang 1986; Kerp 1988, Kerp and Barthel 1993), Lepidopteris Schimper (Townrow 1960), Compsopteris Zalessky (Meyen and Migdissova 1969); Rhapliidopteris Barale (Meyen 1979; Gomankov and Meyen 1986), Tatarina Meyen (Meyen and Gomankov 1980; Gomankov and Meyen 1986; Wang and Wang 1986), and Comia Zalessky (Fefilova 1973). These cuticles are mainly characterized by a poor differentation between the costal and intracostal zones. In general, Callipteris fronds have amphistomatic epidermises, where the intercostal zones consist of irregularly arranged ordinary cells with randomly distributed stomata, and which are separated by narrow and faint costal zones. Hollow and solid papillae and hairs are locally developed. The stomatal apertures are surrounded by a ring of thickened subsidiary cells, which are clearly distinguishable from the ordinary cells, and strong papillae overhang the aperture. Significantly, there are marked differences between the west European Callipteris and the Angaran or so-called ‘Subangaran’ ones, in cell ornamentation: the European Rotliegend species have many elongately hollowed mono- and multicellular papillae (Kerp and Barthel 1993), particularly many multicellular trichomes (Barthel and Haubold 1980, pi. 1, figs 3-5); the Angaran species, in contrast, have numerous solid papillae or thickenings of the periclinal walls (Meyen 1970, pi. 75). This distinction could reflect differences in climate. On the other hand, the Angaran Callipteris are seemingly close to the European Lepidopteris species such as L. martinsii (Kurtze) Townrow, 1960 although their stratigraphical horizons are mostly equivalent to the Kungurian-Kazanian (i.e. so-called ‘Middle Permian’) and are thus higher than the European Rotliegend. Meyen and Migdissova (1969) described poorly preserved cuticles of Compsopteris adzvensis Zalessky from the Upper Permian of the Pechora Basin in West Angara, and this is the only available evidence of the stomatal structure of this genus. The stomata have a monocyclic ring consisting of four to seven subsidiary cells, each of which has a strongly thickened anticlinal wall around the pit as in most callipteroid fronds. In addition, Fefilova (1973) presented some fragments of Comia cuticle, but these are too poorly preserved to identify. The cuticular characters that S. contracta shares with most other peltasperms are its moderate thickness, the less regularly aligned ordinary cells, and the more or less specialized stomatal apparatuses with a prominently thickened rim around the pit, consisting of the raised neighbouring wall of each subsidiary cell. However, the main distinguishing character is the poorly developed ring of subsidiary cells, i.e. most stomata are anomocytic or only partially monocyclic. Among the peltasperms for which cuticular evidence is known, only some Tatarina species have cuticles similar to those of S. contracta. These include species considered to be endemic to the Upper Permian of the so-called ‘Subangara’, such as T. conspicua Meyen (Meyen and Gomankov 1980; EXPLANATION OF PLATE 5 Figs 1-6. Supaia contracta sp. nov. Tianjin Institute of Geology and Mineral Resources; cuticles; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian. 1, 8915-7 (holotype); stoma on lower cuticle; x 400. 2, 9406-sl ; unseparated upper and lower cuticles showing dark-spots caused by fungi; x 40. 3-5, 8915-7 (holotype); lower cuticle; x 40. 4, upper cuticle, showing two or three stomata; x 400. 5, upper cuticle on the rachis; x 200. 6, 8915-1 la; lower cuticle showing rare files of ellongate cell; x 40. PLATE 5 WANG, Supaia 262 PALAEONTOLOGY, VOLUME 40 Gomankov and Meyen 1986). As in S. contracta, they have stomatal apparatuses that are mostly faintly monocyclic and subsidiary cells that are not clearly differentiated from the ordinary ones in size or shape. In addition, a projecting rim surrounding the stomatal aperture is sometimes visible (Gomankov and Meyen 1979, text-figs 7-9) like that of the present new species. Supaia yuanquensis sp. nov. Plate 4, figures 1-2, ?3, 4-9 1955 Protoblechnum wongii Chow et al. , p. 167, pi. 1 [refigured in Gu and Zhi 1974, p. 115, pi. 130, fig. 6 and Liu 1989, p. 449, pi. 6, fig. 1]. 1996 Autunia sp. A Wang, pi. 3, fig. 9. Derivation of name. From Yuanqu district where the type specimens were found. Holotype. TIGM 8860-1 (PI. 4, fig. 1); upper-middle Tianlongsi Formation, Yaotou village, Yuanqu, Shanxi. Diagnosis. Small frond, not exceeding 150 mm long and 100 mm wide. Primary rachis delicate, 1-3 mm thick, forked at acute angle into two equal branches, each one bearing four to six pairs of alternating pinnules. Pinnules narrowly elongate to lanceolate, with a gradually acuminate base and a subacute apex, measuring up to 60 mm long and 20 mm wide. Pinnules acroscopically contracted to form a clear ‘puckering’, and basiscopically decurrent to form a narrow wing along the rachis. Midvein only faint in the decurrent base of the pinnule and then rapidly vanishing. Lateral veins normally sparse but occasionally dense, extending strongly and obliquely from midvein at acute angle (no more than 40°), and bifurcating once or twice. Laminae thinner and weaker in texture than both the above two species. Comparison and remarks. This new species is a form-type with the following features that distinguish it from the other species of the genus : a delicate primary rachis, a smaller frond, narrowly lanceolate pinnules with a decurrent base, a thin pinnule lamina, and sparse, obliquely extending lateral veins. A specimen assigned to Protoblechnum wongii by Chow et al. (1955) from the ‘Upper Shihhotze Formation’ in south-eastern Shanxi shows a forked primary rachis, pinnules with a constricted and decurrent base, and obliquely extending veins. On these features, it is clearly attributable to S', yuanquensis. S. yuanquensis appears to be endemic to south-eastern Shanxi. EXPLANATION OF PLATE 6 Figs 1-5, 10. Supaia contracta sp. nov. Tianjin Institute of Geology and Mineral Resources 9406-SI ; Xuangan Coal Mine, Yuanping, Shanxi; upper Tianlongsi Formation, Upper Permian. 1^4,10, cuticles viewed under SEM. 1, outer side of lower cuticle, showing locally developed papillae; x 180. 2, stoma viewed from inner side; x 900. 3, same; x 1800. 4, inner side of the upper cuticle; x 360. 5, piece of cuticle torn from hand specimen showing slightly fascicled venation with spherical fungal spots along both sides of the veins; x 4. 10, fungal filament attached to cell wall; x 1800. Figs 6-9. Supaia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi ; upper Tianlongsi Formation, Upper Permian. 6, 9306-37 ; minute piece of cuticle showing small spines extending from the corner of cell walls; x 200. 7-8, isolated fungal body. SEM photographs, both from specimen 9306-27 (shown in PI. 2, fig. 5). 7, x 60; 8, details of fig. 7; x 300. 9, details of minute residue of mesophyll; x 180. PLATE 6 WANG, Supaia 264 PALAEONTOLOGY, VOLUME 40 text-fig. 6. Supaia contracta sp. nov. Tianjin Institute of Geology and Mineral Resources; Xuangan Coal Mine, Yuanping, Shanxi; upper Tianlongsi Formation, Upper Permian, a, c, 8915-7 (holotype); main part of bipartite fronds, b, d, 9406-SI; parts of branch, a-b, x 1; c-D, x 2. Family peltaspermaceae Thomas ex Harris, 1937 Form-genus autunia Krasser, 1919 Autunia shanxiensis sp. nov. Text-figs 8, 9A-D 1996 Aff. Autunia sp. Wang, pi. 1, fig. 5; pi. 2, fig. 4. Derivation of name. From Shanxi province where the type specimens were found. WANG: PERMIAN SUPAIA AND AUTUNIA 265 text-fig. 7. Supaia contracta sp. nov. Tianjin Institute of Geology and Mineral Resources; all SEM photographs from specimen 8915-7 (holotype, see Text-fig. 8a); Xuangan Coal Mine, Yuanping, Shanxi; upper Tianlongsi Formation, Upper Permian, a, piece of cuticle from near rachis, showing fungal filament (0; x 300. b, inner side of lower cuticle, showing remains of large fungal body (right); x 180. c, showing a covering membrane of a fungal body; x 180. d, inner side of stoma, showing guard cells infected by fungi; x 900. E, outer side of upper cuticle, showing the raised circular rim of subsidiary cells; x 90. f, showing the degraded cell walls; x 900. G-i, stomata viewed from outer side, showing sunken guard cells and raised rim surrounding the aperture; x 1800. 266 PALAEONTOLOGY, VOLUME 40 text-fig. 8. Autunia shanxiensis sp. nov. Tianjin Institute of Geology and Mineral Resources; Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian, a, 9306-33 (holotype); ovuliferous cone with ovules; many megasporphylls spirally and vertically attached to rachis, probably each one with two reversed ovules; x 2. b-c, e, ovuliferous cones having shed ovules. B, 9306-39; x 1. c, 9306-31; x 1. E, 9306-36, x 2. D, f-g, isolated megaporophylls. d, 9306-32; x 2. f, 9306-32; x 5. G, 9306-58; x 5. WANG: PERMIAN SUPAIA AND AUTUNIA 267 Holotype. TIGM 9306-33; upper Tianlongsi Formation, Wangtao village, Qinyuan, Shanxi. The specimen is illustrated in Text-figure 8a, and is closely associated with many Supaia shanxiensis fronds. Diagnosis. Detached ovuliferous cone, 20-25 mm in diameter and probably over 100 mm long. The cone is cylindrical, abruptly constricted at both ends. Cone axis rather thick, 3-5 mm wide, bearing spirally arranged megasporophylls attached at a large angle or at right-angles. Megasporophyll has a peltate or semi-round head and is bilaterally symmetrical, with five or six lobes on the border and has a funnel-form in the central portion passing immediately into a slender petiole which attaches to the rachis with its prominently expanding base. On the adaxial side of the peltate head, two or probably more flattened seeds or ovules are fixed. Seeds elliptical, Carpolithus- type, 3-5 mm in length, with a narrow border around the nucellus, slightly concave at its bottom and obtusely acute at the apex. Remarks and comparison. Three species were originally attributed by Kerp (1982) to the Autunia ovuliferous genus: A. milleryensis (Renault) Krasser, A. thomasii (Thomas) Kerp and A. dzungarica (Salymenova) Kerp. In addition, he regarded Sandrewia texana Mamay, 1975, from the Lower Permian of Texas and Kansas, as a synonym of A. milleryensis , though its associated callipterid frond had not been accurately identified. Kerp (1988) later successfully documented the direct connection between the Callipteris confer ta frond and the ovuliferous A. milleryensis , both having the same type of epidermal structure, and both being consistently associated in most localites in Europe. Significantly, he utilized Autunia as the name of a natural genus, in which he included two species: A. conferta (Sternberg) Kerp and A. naumannii (Gutbier) Kerp. According to the diagnosis emended by Kerp (1988, p. 305), A. naumannii differs substantially from A. milleryensis , mainly in the sterile frond having wedge-lobed, acroscopically constricted and strongly basiscopically decurrent pinnules, and in the fertile fronds having modified pinnules, but not in the architecture of the ovuliferous organs themselves. Furthermore, when emending the organ-genus (sensu pre-1978 editions of ICBN) Peltaspermum Harris, to form another peltaspermous natural genus, Poort and Kerp (1990) established Meyenopteris as a natural genus for replacing Autunia thomasii and the form-genus Autuniopsis for Autunia dzungarica. As to the nature of the ovules in Autunia , Kerp (1982, 1988) and Meyen (1984) presumed that they were abaxially attached to the megasporophyll, but none of the illustrated photographs clearly indicate their actual attachment. The specimens of A. shanxiensis show no prominent traces or scars where ovules could have been attached to both sides of the petiole, but instead two or probably more scars are clearly visible on the adaxial surface of the petiole head (Text-fig. 8g), as in Peltaspermum. A. shanxiensis exhibits the following characters that distinguish it from the above mentioned Euramerican types: all the ovuliferous megasporophylls were closely imbricated into an independent cone rather than forming a laxly fertile dwarf cone as in A. milleryensis ; its peltate head shows obscured or weak ribs, and has a markedly constricted or cordate base, which contrasts with the gradually accuminate, wedge-formed base of the petiole and the prominent radial ribs on the head seen in the European species. Among previous records, only a fructification assigned by Meyen (1982, text-fig. 18) to Peltaspermum ? sp. A from the Upper Permian of the Pechora Basin can be compared to A. shanxiensis. This is also a discrete cone like A. shanxiensis , especially in lateral view, but the more or less bilateral head of its megasporophyll shows clear radial ribs. Significantly, Meyen's fructification is associated with a Compsopteris rather than a Callipteris frond. (?)Pollen-bearing cone Text-figure 9e-f Description. Only one specimen shows a cone about 20 mm in diameter, with a strong, 3 mm wide rachis to which many oboval scales are spirally attached. Each scale divides distally into two or three small lobes, on which many spherical bodies are adhered as microsporangia. 268 PALAEONTOLOGY, VOLUME 40 text-fig. 9. a-d, Autunia shanxiertsis sp. nov.; parts of cone having shed ovules, a, 9306-57; B-c, 9306-56; d, 9306-40. e-f, apical part of (?)pollen-bearing cone; 9306-55. All from Wangtao village, Qinyuan, Shanxi; upper Tianlongsi Formation, Upper Permian; Tianjin Institue of Geology and Mineral Resourses a, c, x 5; b, d-e, x 2; F, x 6. Remarks. This cone is closely associated with the foregoing Supaia frond and Autunia ovuliferous cone, but unfortunately no details of its microscopic structures are visible, to establish its nature and affinities. DISCUSSION Systematics and nomenclature When establishing the genus Supaia , White (1929, p. 55) interpreted it to be a product of ‘a frond reduction in the direction of simplicity as a result of environmental adaptation to a seini-arid climate, from bipinnate Callipteris to monopinnate Supaia although he also cited some comparison of Supaia with several Permian alethopteroid fronds, such as Protoblechnum from China, Glenopteris Sellards from America, ‘ Odontopteris rossica' Zalessky from Angara, and even some Mesozoic forked pteridosperm fronds, including Dicroidium Gothan, Thinnfeldia Ettingshausen and ‘ Danae- opsis' hughsii Feistmantel. Many palaeobotanists have accepted this view (Neuburg 1948; Asama WANG: PERMIAN SUPAIA AND AUTUNIA 269 1960. 1985; Boureau and Doubinger 1975; Haubold 1980) and it has become generally accepted that the characteristics of Supaici are its bipartite frond with two monopinnate branches, and the occur- rence of several pairs of pinnules on both sides of the basal rachis below the fork. Due to the limited knowledge of its fructifications (White 1934; Mamay and Watt 1971), palaeobotanists have tended not to be concerned with its systematic position, but with the significance of its limited distribution in the USA and to its correlation with allied fronds (Neuburg 1948; Sze 1955a, 19556; Read and Mamay 1964; Chaloner and Meyen 1973; Huang 1977; Lemoigne 1988; Liu 1989). The late Serge Meyen (1984, p. 47, text-fig. 16-3) was the first to classify Supaia in the Peltaspermales, based on a report of pinnate ovuliferous organs associated with Callipteris and Supaia fronds in America (Mamay and Watt 1971), although no detailed information on these fructifications was then available. The obligate association reported in this paper of Supaia fronds and an Autunia ovuliferous cone strongly supports Meyen's classification. Furthermore, special attention should be paid to the similarity in cuticular structure between Supaia contractu from North China and some Tatarina species from the Upper Permian of the Urals, as the latter was also classified by Meyen in the Peltaspermales. In gross morphology, Tatarina foliage (i.e. the form- genus Pursongia Zalessky) shows considerable variation from simple, undissected and bifurcate, to pinnate or pinnatifid fronds like callipterids. This is thus very similar to the trend of reduction of fronds from Callipteris to Supaia recognized by White (1929) in the areas of semi-arid climate, although its associated ovuliferous scales are radially symmetrical similar to those of Peltaspermopsis , and differ markedly from the Autunia- type scales. Significantly, Tatarina has been identified from dispersed cuticles from the uppermost Permian Sunjiagou Formation in Shanxi (Wang and Wang 1986), overlying the present Supaia- bearing bed. It thus seems that an evolutionary series could have occurred in the bipartite fronds in eastern Laurasia during the middle to late Permian, from the bipinnatilid Protoblechnum wongii (Liu 1989) via the monopinnate Supaia to the undissected or monopinnate Tatarina , and is similar to what Asama (1960) referred to as growth retardation. The main difficulty that faces me in the identification of Supaia comes from the complicated and confused nomenclature for pteridosperms generally. The establishment of a fossil natural taxon is a long-term goal sought by current palaeobotanists. Where many of the plant’s organs are preserved in isolation, whole-plant reconstructions should be a prerequisite before establishing such a natural taxon. At present, however, the opportunity for establishing such reconstructions for the fossil gynmosperms of the Permian-Triassic red-beds has not yet arisen, because their fructifications are mostly separated from the parent plants during preservation. Although Kerp (1988) has successfully demonstrated the direct relationship between an Autunia ovuliferous organ and the frond known as Callipteris conferta , it is much more difficult to ascertain exactly how this fructification was attached to the parent plant, and so the organizational architecture of the whole plant is still obscure. We do not know if the ovuliferous organ was an imbricate strobus or only an axis with laxly attached megasporophylls; nor can we ascertain whether the organ should correspond in architecture to an entire frond or only to a pinna or modified pinnule. Of special importance, because of the lack of evidence of the anatomy of the ovuliferous organ, the exact number of ovules in each megasporophyll is not yet even confirmed, although two ovules tend to be reported in most investigations on the gross morphology. The attachment of the ovules is also in question - were they abaxially fixed on both sides of the petiole, or only on the adaxial surface of the disk-like head, as in Peltaspermum ? Such differences would be important enough to distinguish generic or even suprageneric taxa. A natural genus should not be a potentially compound taxon that may have to be divided in the future. Another, purely nomenclatural question arises from the change in usage of a name, initially defined as a form-genus, to being a newly erected natural genus. The emendation or expansion of its original definition will result in a number of specimens that were assigned to the form-genus having to be transferred elsewhere, causing considerable confusion. Based on the obligate association of the ovuliferous Autunia milleryensis and the Callipteris conferta frond in coeval strata at many localities in west-central Europe, Kerp (1988) established a natural genus which he referred to as Autunia. However, he then had to erect a new form-genus, Autuniopsis Poort and 270 PALAEONTOLOGY, VOLUME 40 Kerp, 1990, to replace the old organ-genus (in the sense of the pre-1978 editions of ICBN) concept, although the type specimen (Salymenova 1979) had not then been described in detail. Far more difficulties have arisen in the nomenclature of Peltaspermum. The genus was introduced by Harris (1937) for a type of Late Triassic, pteridospermous ovuliferous organ, closely associated with the frond Lepidopteris otlonis (Goppert) Schimper. Many other similar organs from the Permian-Triassic of Laurasia were later attributed to it (Townrow 1960; Dobruskina 1980; Wang and Wang 1986). Poort and Kerp (1990) changed the essence of the genus from that of a form-genus to that of a natural genus. Following Gomankov and Meyen (1986), they used the name Lopadangium Zhao, which was originally based on an isolated organ of uncertain nature, as a form- genus to receive almost all of the specimens from outside of Europe, that had previously been assigned to Peltaspermum. In fact, most of these specimens had been demonstrated to be peltasperms, either because of the nature of their cuticles or through obligate association; for example, P. usense Dobruskina is associated with Lepidopteris / Scytophyllum fronds in the middle to upper Triassic of the Pechora Basin (Dobruskina 1980, p. 101 ; 1994, p. 307) and P. dafengshanse Wang and Wang is associated with Callipteris and Tatarina fronds in the Upper Permian of Shanxi, China (Wang and Wang 1986). Lopadangium is only a disc-like organ found in the purely Cathaysian floras of South China (Zhao et at. 1980), whose relationship with the peltasperms or even the pteridosperms has not been established. If the exact affinities of the type specimens of Lopadangium can be established in the future, all of those specimens transferred there by Poort and Kerp (1990) will have to be moved again. In practice, there are great difficulties with using natural genera for Permian peltasperms. On the one hand, very similar callipterid foliage is thought to have borne different types of fructification, such as Autunia and Peltaspermum (Naugolnykh and Kerp 1996); on the other, the same type of fructification ( Autunia ) can be found in close association with different frond-types ( Supaia , Callipteris). It is clear that neither Autunia nor Peltaspermum are correlated with a particular type of foliage and so cannot be the bases of natural genera. The Permian-Triassic red-bed floras contain most of the earliest-known gymnosperms that occurred in dry and drier terrains. These gymnosperms are characterized by great species diversity and marked polymorphism in plant ontogeny, which may have been the result of adaptation to a semi-arid or arid climate. They are thus mostly endemic to certain restricted areas. They are also of uncertain affinities due to their isolated distribution and poor preservation, resulting in a lack of knowledge of their fructifications. In nomenclature, the concepts of satellite taxa hierarchy (Thomas and Brack-Hanes, 1984) and operational taxonomic units (Bateman et al. 1992) should be encouraged in such cases as this, i.e. it is better to retain Autunia and Peltaspermum as form-generic names. P alaeophytogeography During the Permian, great changes occurred in the ecosystem, resulting in the development of global biotic provincialism and, in particular, the four main palaeophytogeographical provinces. In the plant kingdom, the considerable development of xeromorphic elements, especially among the seed plants, resulted in these plants dominating the Permian red-bed floras, as they were better adapted to seasonally dry climates. These plants were susceptible to edaphic variation and endemic representives therefore reflect the range of regional physical habihats. As stated above, Supaia is famous for being reputedly restricted to certain areas in western North America and has sometimes been regarded as a characteristic endemic genus of that area (Read and Mamay 1964; Chaloner and Meyen 1973; Meyen in Vakhrameev et al. 1978; Lemoigne 1988; DiMichele and Hook 1992). Although occasionally mentioned in the earlier literature, its abundant occurrence in the Permian of Shanxi reported in the present paper will bring this genus more attention in palaeobotany and especially in palaeophytogeography. Some superficial similarities of the Permian floral components in East Asia to those of the south- western United States produced some discussion and argument in the 1930s (Darrah 1937; WANG: PERMIAN SUPAIA AND AUTUN1A 271 Jongmans and Gothan 1937). These comparisons were largely based on the distribution of enigmatic groups (Sze 1955a, 1955/?; Lemoigne 1988) but more detailed studies in both areas have shown that the similarities are often at supragenric rank, such as between the taeniopterids, callipterids, gigantopterids, noeggerathiopterids and conifers. There is in fact no strong support for a close relationship between the Permian floras of China and North America, in spite of the apparent presence of many common elements, such as Taeniopteris, Callipteris , Prototoblechnum (= Glenopteris ), Sphenopteridium , Discinites , Russellites , Lesley a and Walchia (Mamay 1966, 1968, 1989, 1990, 1992; Mamay et al. 1984, 1988). Regrettably, little attention has been paid to this in recent papers on Permian biogeography (Wagner 1993; Wnuk 1996). The similarity between these floras remains open to question. The Supaia flora of south-western North America is characterized by low diversity: the total number of whole-plant species may be as low as 1 5 to 20. The Supaia plant was also highly xeromorphic, with leathery laminae covered with heavy hairs, spines and sunken veins (DiMichele and Hook 1992). Ziegler (1990) applied the concept of extant biomes to illustrate climatically the Permian phytogeographical world. Western North America (together with Europe) was assigned to the low diversity, summer-wet Biome 2, characterized by the occurrence of Callipteris. On the basis of outdated information. North China was mistakenly placed in the tropical, ever-wet Biome I (i.e. the typical Cathaysian Realm, which also included South China), though the area had been regarded as part of Laurasia during the Pemian and Triassic (Wang 1985) based on the uniform phytostratigraphical sequence. In fact, most of the Permian floral elements of North China are of Eurasian type, except for the gigantopterids, rather than Cathaysian. The Psygmophylluru Zone assemblages bearing Supaia, which can be at least dated to Kungurian-Kazanian (Wang 1996), is typical of a seasonally alternating humid-arid biome, similar to that of the Hermit Shale flora in Texas. Significantly, the Permian floras in south-western North America, including that of the Hermit Shale, are in some aspects very similar to the Psygmophyllum Zone assemblages in North China: all fossil taphocoenoses occur in red-beds; the main components of both are common either at the suprageneric or generic rank; and much of the ecophysiological evidence is the same (this latter issue will be discussed elsewhere). At least in North China, the three species of Supaia show considerable ecomorphic changes from south to north, representing variation in regional physical environments such as habitat heterogeneity and increasing climatic aridity. S. yuanquensis from south Shanxi has relatively small fronds with a slender or delicate primary rachis and more elongate pinnules; its veins extend obliquely from the midrib or rachis; the pinnule lamina is thinner and apparently weakly cutinized. This contrasts with S. contracta from north Shanxi, which has large fronds and pinnules with a thick and leathery lamina and concealed veins; the pinnules have a strongly constricted base and are covered with heavy hairs, papillae and sunken stomata. S. shanxiensis can be regarded as a transitional ecomorph between the other two. The emergence of these three regional ecomorpho- logical types clearly reflects vegetational endemism and isolation, and a regressive succession of Supaia- bearing communities within North China during the Permian (Wang 1993). Ecophysiology Initially, White ( 1 929) presumed Supaia to have belonged to a plant inhabiting areas with a semi-arid or alternating arid-humid climate, with long intervals of desiccation. This was based on a combination of the sedimentary phases of its red-bed matrix, and of the greater simplicity of its compound, pinnate frond. He also noticed that the pinnule lamina was rigid and coriaceous, had concealed veins and was covered with dense scales, and thus probably belonged to an amphibious plant. Particularly, fronds with apparently chewed margins were stated to be the result of the activities of larvae, insects and microbios (nematodes) (White 1929, p. 66). In addition, he mentioned a drainage system in the apical part of the Brongniartites pinnae, explaining it as a 272 PALAEONTOLOGY, VOLUME 40 physiological adaptation for the capture of rain. In this aspect, more important information can be extracted from the present material. Xeromorphic habit. The regressive simplification of the fronds, from the bipinnate Protoblechnum to the monopinnate Supaia , was the result of long-term adaptation to a semi-arid climate. Another indicator of preburial desiccative laminae is the many blanks or gaps left in the laminae when splitting the matrix embedding the frond (Text-fig. 5). Partially rolled or wilted mature fronds reflect air-drying during the growing period of the plant, as in many modern desert plants, i.e. the so-called poikilohydric xerophytes (Brown 1974). In addition, primary rachises with a wilted base and reduced or enrolled pinnules indicate that the plant could have naturally disarticulated due to wilt in the dormant season. A similar case of wilt occurred in some Triassic Isoeies , in which various isolated organs of the same plant are found preserved in parautochthonous burials (Wang 1991). On the other hand, the cuticles show sunken guard-cells below a deep pit with a constricted aperture, which was probably an adaptation to reduce excessive evapo-transpiration (Spicer 1989, p. 324). Taphonomic setting. Most plants of the Permian red-beds in North China were buried in ephemeral water-bodies in an arid-humid alternating terrestrial ecosystem, rather than in a uniform wet- temperate ecosystem with persistent drainage (Wang 1993). As stated above, the Supaia-beanng biostromes are typically parautochthonous burials, representing an association of shore-inhabiting bushes, limited to the terraces surrounding small, seasonal ponds and playas. The size of the pond or playa was roughly equal to the dimensions of the biostrome. I have previously suggested that the gradual regressive fragmentation of the vegetation in North China during the Permian and Triassic is an indicator that the vegetation would gradually disappear (Wang 1993). Stomatal index. Traditionally, cuticular features have tended to be regarded as important evidence for classifying plant taxa and only rarely to be of ecological or ecophysiological significance. However, recent botanical (Murray 1995) and palaeobotanical (Visscher 1993) studies have demonstrated the potential significance of the Stomatal Index and of stomatal aperture size for evaluating the CO, content of palaeoatmospheres and ecophysiological responses to photosynthesis. Such studies started with angiosperms from Recent European mountains and some herbarium specimens from leaves collected over the last 200 years (Woodward 1987). Using fossil leaves, the evidence was then extended back to the Late Tertiary (van der Burgh et a/. 1993; Masterson 1994). In the remote past such as the Permian, however, estimation is very difficult, and can only be based on the relative values of absolute stomatal indices. There exists no model to compare effectively the CO, levels of Permian palaeoatmospheres with that of today. An attempt must be made to accumulate more relevant data, although what the results will yield is far from certain. The stomatal density of Supaia contracta varies from 100-360 mm 2 (mean value 250 mm'2), which is comparable to the values estimated from several xerophytic plants growing today near the margins of the Tungeli Desert in north-western China (Zhao and Huang 1981). The Stomatal Index values from the upper cuticles are mostly 9-1 1, and those of the lower cuticle 13-14 (this is excluding extreme values obtained from cuticles on or near the rachis or margins of the pinnule). Such low values again are indicative of xerophytic plants. Plant-fungal interaction A number of dark spots in or on fossil leaves have been interpreted as various fungal remains since early in the nineteenth century but most were recorded from post-Cretaceous specimens. Knowledge of Palaeozoic epiphyllous fungi, especially the parasitic ones, is very limited. Of the 145 records listed by Dilcher (1965), only 11 are of pre-Cretaceous examples, and most of these are unreliable nineteenth century identifications made prior to the development of microscopic techniques for fungal identification. Thus, Sherwood-Pike (1988) claimed that in the Late WANG: PERMIAN SUPAIA AND AUTUNIA 273 Cretaceous such fungi had not developed to the level that they could be reliably identified, and that the oldest records were from the Palaeocene. However, some fungi had in fact been previously microscopially documented, in Jurassic-Cretaceous cycad leaves (Krassilov 1967 ; Zheng and Zhang 1986) and even in Late Carboniferous plants (Hutchison 1955; Stubblefield et al. 1983). The epiphyllous fungi found within these Permian Supaia fronds are clearly of considerable significance, even though they have yet to be identified in detail. They are the first examples of Palaeozoic leaf-inhabiting fungi with a reliably parasitic relationship, other than the isolated parasitic hyphae recorded in a Late Carboniferous gymnospermous cone (Stubblefield et al. 1984). More importantly, these fungal-spots on the laminae occur regularly along both sides of the veins (PI. 6, fig. 5) all over both the lamina and primary rachis of the frond. This contrasts with the locally isolated or random distribution of such spots in previous records. Numerous fungal colonies growing within the mesophyll would produce little or no damage to the cuticles and the occurrence of small rounded bodies, termed wall appositions or callosities, which are attached to the wall of cell also supports the inference of a parasitic relationship, as argued by Stubblefield et al. (1984) and Taylor and Osborn (1996) based on Devonian-Carboniferous material. In addition, the specialization of the fungal infection unique to this kind of frond can be affirmed at two locations yielding Supaia in Shanxi. These localities are more than 45 km apart and it can therefore be assumed that both represented an ecologic niche where the vegetation was suffering from a geographically widespread infectious disease. Of course, expansion and reduction of an ecotonal vegetation could have been directedly related to the decreasing survival of vegetation towards the end of the Permian. Desert conditions were then starting to develop in North China, as it became part of the vast barren terrain in the Northern Hemisphere, where there were many factors militating severely against vegetation survival (e.g. severe water limitation, strong winds, rainstorms, wild-fire) and which effectively destroyed most of the continental vegetation (Wang 1993). The infectious fungal disease may therefore be seen as part of a general ecological crisis that developed towards the end of the Permian (Eshet et al. 1995; Visscher et al. 1996). Acknowledgements. This project is supported by the National Natural Science Foundation of China (NSFC grant No. 39370050). My greatest thanks go to Dr Christopher J. Cleal (National Museum and Gallery of Wales, Cardiff) for helpful comments on the manuscript and particularly, for effective assistance in linguistics to get it into the present form; I also thank Dr Sergius H. Mamay (Smithsonian Institution, Washington) for his critical reviews and providing valuable references, Mrs Chen An-shu of our Institute for technical assistance and Mrs Zhong Xin-bao for drafting. REFERENCES asama, K. 1960. Evolution of the leaf forms through the ages explained by successive retardation and neoteny. Science Reports of the Tohoku University , Sendai , 2nd Series (Geology Special Volume ), 4, 252-280. — 1985. Permian to Triassic floral change and some problems of the paleobiogeography, parallelism, mixed floras and origin of the angiosperms. 199-208. In nakazawa, k. k. and dickens, j. m. (eds). The Tethys, her paleogeography and paleobiogeography from Paleozoic to Mesozoic. Tokai University Press, Tokyo, 317 pp. barthel, m. 1962. Epidermisuntersuchungen an einigen inkohlten Pteridospermenblattern des Oberkarbons und Perms. Geologie , 33, 1-140. and haubold, h. 1980. Zur Gattung Callipteris Brongniart. Teil I: Die Ausbildung von Callipteris conferta (Sternberg) Brongniart im mitteleuropaischen Rotliegenden. Schriftenreihe fur Geologische Wissenschaft , 16, 49-105. bateman, r. m., dimichele, w. a. and willard, d. a. 1992. Experimental cladistic analysis of anatomically preserved arborescent lycopsids from the Carboniferous of Euramerica: an essay on paleobotanical phylogenetics. Annals of the Missouri Botanical Garden , 79, 500-559. boureau, e. and doubinger, j. 1975. Trade de paleobotanique. Tome IV(2 ), Pteridophvlla. Masson et Cie, Paris, 768 pp. brown, g. w. (ed.) 1974. Desert biology, volume II. Academic Press, New York, 601 pp. burgh, j. van der, visscher, h., dilcher, d. l. and kurschner w. m. 1993. Paleoatmospheric signatures in Neogene fossil leaves. Science , 260, 1788-1790. 274 PALAEONTOLOGY, VOLUME 40 chaloner, w. g. and meyen, s. v. 1973. Carboniferous and Permian floras of the northern continents. 169-186. In hallam, a. (ed. ). Atlas of palaeobiogeography. Elsevier, Amsterdam, 531 pp. chen han-qing and niu ying-xue 1993. [Division and correlation of multilayer strata of Upper Palaeozic in Xishan, Taiyuan], Shanxi Geology, 8, 15-20. [In Chinese], chow tse-yen (= Zhou Zhi-yan), chang shan-chen and chang lu-jin 1955. An additional specimen of forked frond of Protoblechnum wongii Halle. Acta Palaeotologica Sinica, 3, 167-171. [In Chinese with English summary], cleal, c. J. and thomas, b. a. 1991. Carboniferous and Permian palaeogeography. 154-181. In cleal, c. j. (ed.). Plant fossils in geological investigation ', the Palaeozoic. Ellis Horwood, London, 233 pp. darrah, w. c. 1937. Some floral relationships between the late Paleozoic of Asia and North America. Problem y Paleontologii , 5, 195-205. dijkstra, s. J. 1965. Supaia D. White. 3356-3358. In dijkstra, s. j. (ed.). Fossil cat alogus, II. Plantae, 62. Junk, s-Gravenhage, 3309-3420. dilcher, d. l. 1965. Epiphyllous fungi from Eocene deposits in Western Tennessee, U.S.A. Palaeontographica , Abteilung B , 116, 1-54. dimichele, w. a. and hook, r. w. 1992. Paleozoic terrestrial ecosystems. 205-326. In behrensmeyer, a. k., damuth, J. D., dimichele, w. A., potts, R., SUES, H.-D. and wing, s. L. (eds). Terrestrial ecosystems through time ', evolutionary paleoecology of terrestrial plants and animals. The University of Chicago Press, Chicago and London, 568 pp. dobruskina, i. a. 1980. Stratigraficheskoe polozhenie floronosnykh tolshch triasa Evrazii. Trudy Geologicheskii Institut, Akademiya Nauk SSSR, 346, 1-161. [In Russian], 1994. Triassic of Eurasia. Schriften der Erdwissenschafts Kommission, Osterreicher Akademie fur Wissenschaft , 10, 1-422. doubinger, J. and heyler, d. 1975. Nouveaux fossiles dans le Permien frangais. Bulletin de la Societe Geologique de France, 7e serie, 17, 1176-1180. eshet, Y., rampino, m. r. and visscher, h. 1995. Fungal event and palynological record of ecological crisis and recovery across the Permian-Triassic boundary. Geology, 23, 967-970. fefilova, l. a. 1973. Paporotnikovidnye permi sebera PreduraT skogo progiba. Nauka, Leningrad, 192 pp. [In Russian]. gomankov, a. v. and meyen, s. v. 1979. O predstavitelyakh semelstva Peltaspermaceae iz permskikh otlozhenil Russkoi platformy. Paleontologicheskii Zhurnal, 1979, 124—138. [In Russian]. - 1986. Tatarinovaya flora (sostav i rasprostranenie v pozdnei permi Evrazii). Trudy Geologicheskii Institut, Akademiya Nauk SSSR, 401, 1-174. [In Russian]. Harris, t. m. 1937. The fossil flora of Scoresby Sound, East Greenland. Part 5: Stratigraphic relations of the plant beds. Meddeleser om Gronland, 113, 1-112. haubold, h. 1980. Zur Gattung Callipteris Brongniart. Teil II. Indizien fur nordamerikanische Callipteriden - Tinsleya, Yakia und Supaia - im Rotliegenden des Thiiringer Waldes und der Innersudetischen Mulde. Zeitschrift fur Geologische Wissenschaft . 8, 747-767. HUANG BEN-HONG 1977. [Permian flora from the south-eastern Part of the Xiao Hinggan-lin (Lesser Khingan), NE China], Geological Publishing House, Peking, 79 pp. [In Chinese]. hutchinson, s. a. 1955. A review of the genus Sporocarpon Williamson. Annals of Botany, Series 2, 69, 425-437. INSTITUTE OF GEOLOGY AND PALAEONTOLOGY AND INSTITUTE OF BOTANY, ACADEMIA SINICA (‘GU and ZHl’)1974. [Fossil plants of China I. Palaeozoic plants from China]. Science Press, Beijing, 277 pp. [In Chinese]. jongmans, w. j. and gothan, w. 1937. Betrachtungen iiber die Ergebnisse des Zweiten Kongresses fiir Karbonstratigraphie. Deuxieme Congres pour T Avancement des Etudes de Stratigraphie Carbonifere, Compte rendu (Heerlen, 1935), 1, 1-40. kerp, j. h. f. 1982. Aspects of Permian paleobotany and palynology. II. On the presence of the ovuliferous organ Autunia milleryensis (Renault) Krasser (Peltaspermaceae) in the Lower Permian of the Nahe area (F. G. R.) and its relationship to Callipteris conferta (Sternberg) Brongniart. Acta Botanica Neerlandica, 31, 417-427. 1988. Aspects of Permian paleobotany and palynology. X. The west and central European species of the genus Autunia Krasser emend. Kerp (Peltaspermaceae) and form-genus Rhachiphyllum Kerp (callipterid foliage). Review of Palaeobotany and Palynology , 54, 249-360. and barthel, m. 1993. Problems of cuticular analysis of pteridosperms. Review of Palaeobotany and Palynology, 78, 1-18. krasser, f. 1919. Studien iiber die fertile Region der Cycadophyten aus dem Lunzer Schichten: WANG: PERMIAN SUPAIA AND AUTUNIA 275 Makrosporophylle. Denkschriften der Kaiserlichen Akademie der Wissenschaften , Wien , Mathematisch- Naturwissenschaften Klasse , 97, 1-32. krassilov, v. A. 1967. Rcmnemelovaya flora yuzhnogo Primorya i ego znachenie dlya stratigrafii. Nauka, Moscow, 264 pp. [In Russian]. lemoigne, y. 1988. La flore au cours des temp geologigues. Tome 1. Geobios , Memoire Special , 10, 1-384. Liu lu-jun 1989. Discovery of bipinnatifid fronds of Protoblechnum wongii Halle with discussion. Acta Palaeontologica Sinica , 28, 447-454. [In Chinese and English]. mamay, s. h. 1966. Tinsleya , a new genus of seed-bearing callipterid plants from the Permian of north-central Texas. Professional Papers of the United States Geological Survey , 523-E, 1-14. 1968. Russellites, new genus, a problematical plant from the Lower Permian of Texas. Professional Papers of the Unites States Geological Survey , 593-1, 1-13. 1975. Sandrewia , n. gen., a problemtical plant from the Lower Permian of Texas and Kansas. Review of Palaeobotany and Palynology , 20, 75-83. — 1989. Evolsonia , a new genus of Gigantopteridaceae from the Lower Permian Vale Formation, north- Ccntral Texas. American Journal of Botany , 76, 1299-131 1. — 1990. Charliea manzanitana, n. gen., n. sp., and other enigmatic parallel-veined foliar forms from the Upper Pennsylvanian of New Mexico and Texas. American Journal of Botany, 77, 858-866. 1992. Sphenopteridium and Telangiopsis in a Diplopteridium- like association from the Virgilian (Upper Pennsylvanian) of New Mexico. American Journal of Botany, 79, 1092-1101. 1995. Reinstatement of the fossil name Russellites (not a synonym of Yuania). Taxon, 44, 43-51. — miller j. M. and rohr, D. M. 1984. Late Leonardian plants from west Texas: the youngest Paleozoic plant megafossils in North America. Science, 223, 279-281. — and stein Jr, w. E. 1988. Foliar morphology and anatomy of the gigantopterid plant Delnortea abbottiae, from the Lower Permian of West Texas. American Journal of Botany, 75, 1409-1433. — and watt, a. D. 1971. An ovuliferoid plant frond from the Hermit Shale (Lower Permian) of the Grand Arizona. Professional Papers of the United States Geological Survey, 750-C, 48-51. masterson, J. 1994. Stomatal size in fossil plants: evidence for polyploidy in majority of angiosperms. Science, 264. 421-124. meyen, s. v. 1970. Epidermisuntersuchungen an permischen Landpflanzen des Angaragebietes. Palaontolo- gische Abhandlungen , Abteilung B, 3, 523-552. 1979. Permian predecessors of the Mesozoic pteridosperms in western Angaraland, U.S.S.R. Review of Palaeobotany and Palynology, 28, 191-201. 1982. The Carboniferous and Permian floras of Angaraland (a synthesis). Biological Memoirs, 7, 1-109. 1984. Basis features of gymnosperm systematics and phylogeny as shown by the fossil record. Botanical Reviews, 50, 1-1 1 1 . 1987. Fundamentals of palaeobotany. Chapman and Hall, London, 432 pp. — and gomankov, a. v. 1980. Pel’taspermovye pteridospermy roda Tatarina. Paleontologicheskii Zhurnal, 1980. 116-132. [In Russian], — and migdissova, a. v. 1969. Epidermal’noe issledovanie angarskikh Callipteris i Compsopteris. 59-84. In meyen, s. v. (ed.). Pteridospermy verkhnego paleozoya i mezozoya. Trudy Geologicheskii Institut, Akademiya Nauk SSSR, 190, 1-107. [In Russian], Murray, d. r. 1995. Plant response to carbon dioxide. American Journal of Botany, 82, 690-697. naugolnykh, s. v. and kerp, H. 1996. Aspects of Permian palaeobotany and palynology. XV. On the oldest known peltasperms with radially symmetrical ovuliferous discs from the Kungurian (uppermost lower Permian) of the Fore-Urals (Russia). Review of Palaeobotany and Palynology, 91, 35-62. nemejc, F. 1968. Palaeobotanika , III. Academia, Prague, 478 pp. — and augusta, J. 1937. Deux nouvelles decouvertes de plantes Permien Moravie. Spisy Vyddvene Pfirodovedeckou Fakultou Karlovy University, Praha, 151, 7-10. neuburg, m. f. 1948. Verkhnepaleozolskaya flora Kuznetskogo basselna. Paleontologiya SSSR , 12, 1-342. [In Russian], norin, e. 1924. The lithological character of the Permian sediments of the Angara Series in Central Shanxi, N. China. Geologiska Foreningens i Stockholm Forhandlingar , 46, 19-55. poort, R. j. and kerp, j. h. f. 1990. Aspects of Permian palaeobotany and palynology. XI. On the recognition of true peltasperms in the Upper Permian of western and central Europe and a reclassification of species formerly included in Peltaspermum Harris. Review of Palaeobotany and Palynology, 63, 197-225. read, c. b. and mamay, s. H. 1964. Upper Paleozoic floral zones and floral provinces of the United States. Professional Papers of the United States Geological Survey, 454-K, 1-35. 276 PALAEONTOLOGY, VOLUME 40 salmenova, K. z. 1979. Osobennosti Permskoi flori yzhnogo Kazakhstana i ee svyazi s sosednimi tlorami. Paleontologichesk.il Zhurnal, 4, 119-127. [In Russian]. sherwood-pike, m. 1988. Fossil leaf-inhabiting fungi from northern Idaho and their ecological significance. Mycologia, 80, 14—22. si xing-jian ( = sze hsin-chine) 1989. Late Palaeozic plants from the Qingshuihe region of Inner Mongolia and the Hequ District of northwestern Shanxi. Palaeontologica Sinica, New Series, All, 1-268. [In Chinese and English]. spicer, r. a. 1989. Physiological characteristics of land plants in relation to environment through time. Transactions of the Royal Society of Edinburgh : Earth Sciences , 80, 321-329. Stubblefield, s. p., taylor, T. n., miller, c. e. and cole G. t. 1983. Studies of Carboniferous fungi. II. The structure and organization of Mycocarpon , Sporocarpon, Dubiocarpon and Coleocarpon (Ascomytina). American Journal of Botany, 70. 1482-1498. 1984. Studies of Paleozoic fungi. III. Fungal parasitism in a Pennsylvanian Gymnosperm. American Journal of Botany, 71, 1275-1282. SZE, H. c. 1955a. On a forked frond of Protoblechnum wongii Halle. Acta Palaeontologica Sinica, 3, 1 1-24. [In Chinese and English]. - 19556. On a forked frond of Protoblechnum wongii Halle. Scientia Sinica, 4, 201-212. taylor, t. n. 1993. The biology and evolution of fossil plants. Prentice Hall, New Jersey, 981 pp. - and osborn, j. m. 1996. The importance of fungi in shaping the paleoecosystem. Review of Palaeobotany and Palynology , 90, 249-262. thomas, b. a. and brack-hanes, s. D. 1984. A new approach to family grouping in the lycophytes. Taxon, 33, 247-255. townrow, j. a. 1960. The Peltaspermaceae, a pteridosperm family of Permian and Triassic age. Palaeontology, 3, 333-361. Vakhrameev, v. a., dobruskina, I. A., meyen, s. v. and zaklinskaya, E. d. 1978. Paldozische und mesozoische Floren Eurasiens und die Phytogeographie dieser Zeit. Gustav Fischer, Jena, 269 pp. visscher, h. 1993. Links with the past in the plant world: cuticles as recorders of diversity, kerogen formation and palaeoatmosphere C02-level. The Palaeobotanist, 42, 86-92. brinkhuis, h., dilcher, d. l., elsik, w. c., eshet, y., looy, c. V., RAMPINO, M. R. and TRAVERSE, A. 1996. The terminal Paleozoic fungal event: evidence of terrestrial ecosystem destabilization and collapse. Proceedings of the National Academy of Science, 93, 2155-2158. vladiminovich, v. p 1986. Vysshie rasteniya. 32-37, pis 135-155. In gorsky, v. p. and kalmykova, m. a. (eds). Atlas kharakternyk kompleksov permskoi fauny i flory Urala i Russkol platformy. Vsesouznyi Nauchno- issledovatelyskil Geologicheskil Instityt (VSEGEH), 328 pp. [In Russian]. wagner, r. h. 1993. Climatic significance of the major chronostratigraphic units of the Upper Palaeozoic. Compte Rendu 1 2e Congres International de Stratigraphie et de Geologie du Carbonifere et Permien, Buenos Aires, 1, 83-108. wang zi-qiang 1985. Palaeovegetion and plate tectonics: palaeophytogeography of North China during Permian and Triassic times. Palaeogeography, Palaeoclimatology, Palaeoecology, 49, 25—45. 1991. Advances on the Permo-Triassic lycopods in N. China. I. An Isoetes from the Mid-Triassic in Northern Shaanxi Province. Palaeontograpliica, Abteilung B, 222, 1-30. - 1992. Successional trends of plant community during the Paleo-Mesophytic transition in North China. Chinese Science Bulletin, 31, 1098-1103. - 1993. Evolutionary ecosystem of Permian-Triassic redbeds in N. China: a historcal record of natural global desertification. 471^176. In lucas, s. g. and morales, m. (eds). The nonmarine Triassic. Bulletin of the New Mexico Museum of Natural History and Science, 3, 1-536. 1996. Past global floristic changes: the Great Eurasian Floral Interchange. Palaeontology, 39, 189-217. - and wang li-xin 1986. [Late Permian fossil plants from the lower part of the Shiqangfeng (Shichienfeng) Group in North China], Bulletin of the Tianjin Institute of Geology and Mineral Resources, Chinese Academy of Sciences , 15, 1-120. [In Chinese], white, d. 1929. Flora of the Hermit Shale in the Grand Canyon, Arizona. Publications of the Carnegie Institution, Washington, 405, 1-221. - 1934. The seeds of Supaia, a Permian pteridosperm (abstract). Science, New Series, 79, 462. wnuk, c. 1996. The development of floristic provinciality during the Middle and Late Paleozoic. Review of Palaeobotany and Palynology , 90, 5-40. woodward, f. i. 1987. Stomatal numbers are sensitive to increases in CO., from pre-industrial levels. Nature, 327, 617-618. WANG: PERMIAN SUPAIA AND AUTUN1A 277 Zalessky, m. d. and tchirkova, h. t. 1935. Observations sur quelques vegetaux fossiles du terrain Permien du bassin de Kousnetzk. Izvestiya Akademii Nauk SSSR, Otdetenie Matematicheskii i Estestvu Nauk , 8-9, 1091-1116. zhao cui-xian and huang zi-shen 1981. [A preliminary study of xeromorphism of some important xerophytes growing in Tungeli Desert], Acta Botanica Sinica, 23, 278-283. [In Chinese]. zhao xiu-hu, MO zhuangguan, zhang shan-zhen and yao zhao-qi 1980. [Late Permian flora from W. Quizhou and E. Yunnan]. 70-122. In nanjing institute of geology and palaeontology, academica sinica (ed.). [ Stratigraphy and palaeontology of the Upper Permian coal-bearing formations in W. Quizhou and E. Yunnan.] Science Press, Beijing, 122 pp. [In Chinese]. zheng shao-lin and zhang wu 1986. [The cuticles of two fossil cycads and epiphytic fungi]. Acta Botanica Sinica , 28, 427-436. [In Chinese]. zhou tong-shun and zhou hui-qin 1986. Fossil plants. 39-69. In institute of geology, Chinese academy of geological sciences and xinjiang bureau of geology and mineral resources (eds). Permian and Triassic strata and fossil assemblages in the Dalongkou area of Jimsar, Xinjiang. Geological Publishing House, Beijing, 262 pp. [In Chinese with English summary]. ziegler, a. m. 1990. Phytogeographic patterns and continental configurations during the Permian Period. 363-379. In mckerrow, w. s. and scotese, c. r. (eds). Palaeozoic palaeogeography and biogeography. Memoir of the Geological Society of London , 12, 1 — 435. WANG ZI-Q1ANG Tianjin Institute of Geology and Mineral Resources Chinese Academy of Geological Sciences Typescript received 28 November 1995 No. 4, 8th Road, Dazhigu, Tianjin 300170 Revised typescript received 22 March 1996 Peoples Republic of China APPENDIX 1: PSYGMO PHYLLU M ZONE FLORAS OF SHANXI A total of 46 species belonging to 36 genera has been identified from the Psygmophyllum Zone in a north-south series of ten sections in Shanxi. The following species are unique for the zone: Psygmophyllum multipartitum Halle, Nystroemia pectiniformis Hall, Rhipidopsis sp., Supaia shanxiensis sp. nov., S. contracta sp. nov., S. yuanquensis sp. nov., Callipterisl laceratifolia Halle, Autunia shanxiensis sp. nov., Peltaspernmm sp., Fascipteridium ellipticum Zhang and Mo, Pelourdea reflexa Halle, Lesleya sp., Ginkgophytopsis sp., Sphenobaiera tenuistriata (Halle) Florin and Lixotheca ( Cladophlebis ) permica (Lee and Wang) Yao and Liu. The following already occur in the underlying Plagiozamites and Chiropteris Zones but may persist into the Psygmophyllum Zone : Lobatannularia ensifolia (Halle) Halle, L. heianensis (Kodaira) Kawasaki, L. lingulata (Halle) Halle, Discinites orientalis Gu and Zhi, Yuania chinensis Du and Zhu ( = Russel/ites chinensis Mamay, 1995), Giantonoclea hallei (Asama) Gu and Zhi, G. lagrelii (Halle) Koidzumi, Callipteris changii Sze, Fascipteris hallei (Kawasaki) Gu and Zhi, F. sinensis (Stockmans and Mathieu) Gu and Zhi, Rhomboidopteris yongwolensis (Kawasaki) Sze, Neuropteridium coreanicum Koiwai, Chiropteris reniformis Kawasaki, Protoblechnum wongii Halle, Taeniopteris taiyuanensis Halle, T. tingii Halle, I . densissima Halle, Norinia sp. and Walchia sp. In addition, there are some relicts of the Carboniferous genera such as Calamites, Annularia , Asterophy llites, Sphenophyllum , Lepidodendron , Cordaites and fern-like plants attributed to form-genera Pecopteris, Sphenopteris and Odontopteris , among others. The most remarkable character of the zone is the predominance of Mesophytic gymnosperms, and the zone is referred to as the oldest gymnospermous assemblage in North China (Wang 1993). In particular, psygmophylloids ( Psygmophyllum , Rhipidopsis, Chiropteris) are unique to this zone. Also important are peltasperms (callipterid fronds, Autunia- and Peltaspermum- type ovuliferous organs) and other allied genera ( Supaia , Protoblechnum-Compsopteris, Neuropteridium, Fascipteridium (similar to Comm- type fronds)). Cycad leaves and megasporophylls, although known from the basal Permian of China, significantly increase diversity in the Psygmophyllum Zone. They include a series of Taeniopteris leaves varying from small, narrow, linear forms to large, broad or elliptical ones. Cycad ovuliferous scales range from the fan-like Norinia to the pinnate Tianbaolinia. In Shanxi Province, ancestors of ginkgos (Sphenobaiera and Ginkgophytopsis ) and conifers ( Walchia) rarely appear in this zone, but are abundant in other areas such as Inner Mongolia ( Ginkgophytonl spirata Shi) and Gansu (conifer Majonica). NOTES FOR AUTHORS The journal Palaeontology is devoted to the publication of papers on all aspects of palaeontology. Review articles are particularly welcome, and short papers can often be published rapidly. A high standard of illustration is a feature of the journal. Four parts are published each year and are sent free to all members of the Association. Typescripts should conform in style to those already published in this journal, and should be sent (with a disk, if possible) to the Secretary of the Publications Committee, Dr R. M. Owens, Department of Geology, National Museum of Wales, Cardiff CF1 3NP, UK, who will supply detailed instructions for authors on request (these are published in Palaeontology 1996, 39, 1065-1075). Special Papers in Palaeontology is a series of substantial separate works conforming to the style of Palaeontology. SPECIAL PAPERS IN PALAEONTOLOGY In addition to publishing Palaeontology the Association also publishes Special Papers in Palaeontology. Members may subscribe to this by writing to the Membership Treasurer: the subscription rate for 1997 is £55 00 (U.S. $120) for Institutional Members, and £20 00 (U.S. $36) for Ordinary and Student Members. A single copy of each Special Paper is available on a non-subscription basis to Ordinary and Student Members only , for their personal use, at a discount of 25 per cent, below the listed prices : contact the Marketing Manager. Non-members may obtain Nos 35-55 (excluding 44) at cover price from Blackwell Publishers Journals, P.O. Box 805, 108 Cowley Road, Oxford OX4 1FH, UK, and older issues from the Marketing Manager. For all orders of Special Papers through the Marketing Manager, please add £1 -50 (U.S. $3) per item for postage and packing. PALAEONTOLOGICAL ASSOCIATION PUBLICATIONS Special Papers in Palaeontology For full catalogue and price list, send a self-addressed, stamped A4 envelope to the Marketing Manager. Numbers 2-46, excluding 44, are still in print and are available together with those listed below : 47. (for 1992): The skeletal morphology, systematics and evolutionary history of holothurians, by p. m. Gilliland. 147 pp., 19 text-figs , 1 1 plates. Price £50 (U.S. $100). 48. (for 1992): Contributions to acritarch and chitinozoan research. Edited by s. g. molyneux and k. j. dorning. 139 pp., 28 text-figs , 22 plates. Price £40 (U.S. $80). 49. (for 1993): Studies in palaeobotany and palynology in honour of Professor W. G. Chaloner, F.R.S. Edited by m. e. collinson and a. c. scott. 187 pp., 38 text-figs, 27 plates. Price £50 (U.S. $100). 50. (for 1993): Turanian ammonite faunas from central Tunisia, by g. r. chancellor, w. j. Kennedy and j. m. Hancock. 118 pp. , 19 text-figs, 37 plates. Price £40 (U.S. $80). 51. (for 1994): Belemnitella from the Upper Campanian and Lower Maastrichtian Chalk of Norfolk, England, by w. k. Christensen. 84 pp., 22 text-figs, 9 plates. Price £35 (U.S. $70). 52. (for 1994): Studies on Carboniferous and Permian vertebrates. Edited by A. r. milner. 148 pp., 51 text-figs, 9 plates. Price £45 (U.S. $90). 53. (for 1995) : Mid-Dinantian ammonoids from the Craven Basin, north-west England, byn.i. riley. 87 pp., 5 1 text-figs, 8 plates. Price £40 (U.S. $80). 54. (for 1995): Taxonomy and evolution of Llandovery biserial graptoloids from the southern Urals, western Kazakhstan, by T. N. koren’ and r. b. rickards. 103 pp., 23 text-figs, 14 plates. Price £40 (U.S. $80). 55. (for 1996): Studies on early land plant spores from Britain. Edited by c. J. cleal. 145 pp., 23 text-figs, 28 plates. Price £45 (U.S. $90). Field Guides to Fossils and Other Publications These are available only from the Marketing Manager. Please add £1-00 (U.S. $2) per book for postage and packing plus £1-50 (U.S. $3) for airmail. Payments should be in Sterling or in U.S. dollars, with all exchange charges prepaid. Cheques should be made payable to the Palaeontological Association. 1. (1983): Fossil Plants of the London Clay, by m. e. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $16) (Members £6 or U.S. $12). 2. (1987): Fossils of the Chalk, compiled by E. owen; edited by a. b. smith. 306 pp., 59 plates. Price £11-50 (U.S. $23) (Members £9-90 or U.S. $20). 3. (1988): Zechstein Reef fossils and their palaeoecology, by n. hollingworth and t. Pettigrew, iv+75 pp. Price £4-95 (U.S. $10) (Members £3-75 or U.S. $7-50). 4. (1991): Fossils of the Oxford Clay, edited by d. m. martill and j. d. Hudson. 286 pp., 44 plates. Price £15 (U.S. $30) (Members £12 or U.S. $24). 5. (1993): Fossils of the Santana and Crato Formations, Brazil, by d. m. martill. 159 pp., 24 plates. Price £10 (U.S. $20) (Members £7.50 or U.S. $15). 6. (1994): Plant fossils of the British Coal Measures, by c. J. cleal and b. a. thomas. 222 pp., 29 plates. Price £12 (U.S. $24) (Members £9 or U.S. $18). 7. (1996): Fossils of the upper Ordovician, edited by D. A. T. harper and a. w. owen. 312 pp., 52 plates. Price £16 (U.S. $32) (Members £12 or U.S. $24). 1985. Atlas of Invertebrate Macrofossils. Edited by j. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii +241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24-95. © The Palaeontological Association, 1997 Palaeontology VOLUME 40 • PART 1 CONTENTS The Permian coral Numidiaphyllum : new insights into anthozoan phylogeny and Triassic scleractinian origins Y. ezaki 1 Upper Ordovician conodonts from the Kalkbank Limestone of Thuringia, Germany ANN ALISA FERRETTI and CHRISTOPHER R. BARNES 15 Mid Mesozoic floras and climates R. N. L. B. HUBBARD and M. C. BOULTER 43 The molluscan periostracum : an important constraint in bivalve evolution ELIZABETH M. HARPER 71 The Jurassic ammonite image database ‘Ammon’ BO LIANG and PAUL L. SMITH 99 Functional significance of the spines of the Ordovician lingulate brachiopod Acanthambonia ANTHONY D. WRIGHT and JAAK NOLVAK 113 A giraffid from the Middle Miocene of the island of Chios, Greece LOUIS DE BONIS, GEORGE D. KOUFOS flflJSEVKET SEN 121 A new pliosaur from the Bajocian of the Neuquen Basin, Argentina ZULMA G ASPARINI 135 The dicynodont Lystrosaurus from the Upper Permian of Zambia : evolutionary and stratigraphical implications G. M. king and I. JENKINS 149 First record of footprints of terrestrial vertebrates from the Upper Permian of the Cis-Urals, Russia VALENTIN P. TVERDOKHLEBOV, GALINA I. TVERDOKHLEBOV A, MICHAEL J. BENTON and GLENN W. STORRS 157 Lower Cambrian cambroclaves ( Incertae sedis) from Xinjiang, China, with comments on the morphological variability of sclerites S. CONWAY MORRIS, J. S. CRAMPTON, XIAO BING and a. j. chapman 167 The morphology and shell microstructure of the thecideidine brachiopod Ancorellina ageri from the Lower Jurassic of Argentina PETER G. BAKER and MIGUEL O. MANCENI DO 191 Two Devonian mitrates from South Africa M. RUT A and J. N. THERON 201 Permian Supaia fronds and an associated Autunia fructification from Shanxi, China WANG ZI-QIANG 245 Printed in Great Britain at the University Press, Cambridge ISSN 0031-0239 V-^a Published by The Palaeontological Association - London Price £45-00 THE PALAEONTOLOGICAL ASSOCIATION (Registered Charity No. 276369) The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1997-1998 President : Professor D. Edwards F.R.S., Department of Earth Sciences, University of Wales College of Cardiff, Cardiff CF1 3 YE Vice-Presidents: Dr P. D. Lane, Department of Earth Sciences, University of Keele, Keele, Staffordshire ST5 5BG Dr P. Doyle, Department of Earth Sciences, University of Greenwich, Grenville Building, Pembroke, Chatham Maritime, Kent ME4 4AW Treasurer : Dr T. J. Palmer, Institute of Earth Studies, University of Wales, Aberystwyth, Dyfed SY23 3DB Membership Treasurer : Dr M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL Institutional Membership Treasurer : Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ Secretary : Dr M. P. Smith, School of Earth Sciences, University of Birmingham, Birmingham B15 2TT Newsletter Editor: Dr S. Rigby, Department of Geology and Geophysics, Grant Institute, West Mains Road, Edinburgh EH9 3JW (co-opted) Marketing Manager : Dr A. King, English Nature, Northminster House, Peterborough PEI 1UA Publicity Officer: Dr M. A. Purnell, Department of Geology, University of Leicester, University Road, Leicester LEI 7RH Editors Dr D. M. Unwin, Geology Department, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ Dr R. Wood, Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ Dr D. A. T. Harper, Department of Geology, University College, Galway, Ireland Dr A. R. Hemsley, Department of Earth Sciences, University of Wales College of Cardiff, Cardiff CF1 3 YE Dr R. M. Owens, Department of Geology, National Museum of Wales, Cardiff CF1 3NP Dr B. M. Cox, British Geological Survey, Keyworth, Nottingham NG12 5GG Dr D. K. Loydell (Technical Editor), Department of Geology, University of Portsmouth, Burnaby Building, Burnaby Road, Portsmouth POl 3QL Other Members : Dr P. Pearson, Geology Department, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ Dr M. J. Simms, Department of Geology, Ulster Museum, Botanic Gardens, Belfast BT9 5AB Mr F. W. J. Bryant, 27, The Crescent, Maidenhead, Berkshire SL6 6AA Overseas Representatives Argentina : Dr M. O. Mancenido, Division Paleozoologia invertebrados, Facultad de Ciencias Naturales y Museo, Paseo del Bosque, 1900 La Plata. Australia : Dr K. J. McNamara, Western Australian Museum, Francis Street, Perth, Western Australia 6000. Canada : Professor S. H. Williams, Department of Earth Sciences, Memorial University, St John’s, Newfoundland A 1 B 3X5. China : Dr Chang Mee-mann, Institute of Vertebrate Palaeontology and Palaeoanthropology, Academia Sinica, P.O. Box 643, Beijing. Dr Rong Jia-yu, Nanjing Institute of Geology and Palaeontology, Chi-Ming-Ssu, Nanjing. France: Dr J.-L. Henry, Institut de Geologie, Universite de Rennes, Campus de Beaulieu, Avenue du General Leclerc, 35042 Rennes Cedex. Iberia: Professor F. Alvarez, Departamento de Geologia, Universidad de Oviedo, CL Jesus Arias de Velasco, s/n. 33005 Oviedo, Spain. Japan: Dr I. Hayami, University Museum, University of Tokyo, Hongo 7-3-1, Tokyo. New Zealand: DrR. A. Cooper, New Zealand Geological Survey, P.O. Box 30368, Lower Hutt. Scandinavia : Dr R. Bromley, Fredskovvej 4, 2840 Holte, Denmark. USA : Professor A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66044. Professor N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403. Professor M. A. Wilson, Department of Geology, College of Wooster, Wooster, Ohio 44961. Germany : Professor F. T. Fursich, Institut fur Palaontologie, Universitat, D8700 Wurzburg, Pleicherwall 1 MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1997 are : Institutional membership . . £90 00 (U.S. $175) Student membership .... £1 1-50 (U.S. $20) Ordinary membership £28 00 (U.S. $50) Retired membership .... £14 00 (U.S. $25) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ. Student members are persons receiving full-time instruction at educational institutions recognized by the Council. On first applying for membership, an application form should be obtained from the Membership Treasurer: Dr M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1997 will receive Palaeontology. Volume 40, Parts 1-4. Enquiries concerning back numbers should be directed to the Marketing Manager. Non-members may subscribe, and also obtain back issues up to five years old, at cover price through Blackwell Publishers Journals, P.O. Box 805, 108 Cowley Road, Oxford OX4 1FH, UK. For older issues contact the Marketing Manager. US Mailing: Second class postage paid at Rahway, New Jersey. Postmaster: send address corrections to Palaeontology, c/o Mercury Airfreight International Ltd, 2323 EF Randolph Avenue, Avenel, NJ 07001, USA (US mailing agent). Cover: SEM of sporangium of Cooksonia pertoni ssp. apiculispora from Brown Clee Hill, Shropshire. This is the specimen (NMW 94 60G) that confirmed the vascular status of Cooksonia ; x 70. Photograph published originally in Nature, 357, 683-685, figure la. s p H < oi Oh W Ch-h d> O C/5 m 43 __ JS WJ c3 E. CX Uh C O c£ .C 2 o 4-h aj Oh G OJ 1— C 43 H *" tH 3 >. cl 40 ^ r~ <1> 1- 5 c *a L « 8 g -> c* ^3 H .S C/5 « y < £ POSTCRAN I AL MORPHOLOGY AND LOCOMOTOR BEHAVIOUR OF TWO EARLY EOCENE MIACOID CARNIVORANS, VULPA VUS AND DIDYMICTIS by RONALD E. HEINRICH and KENNETH D. ROSE Abstract. The postcranial skeletons of two contemporaneous early Eocene carnivorans, the miacid Vulpavus and the viverravid Didymictis , are described and compared with behaviourally diverse small and medium- bodied extant carnivorans. Body mass estimates based on the cross sectional geometry of humeri and femora indicate that these two taxa were similar in size, estimates for both genera ranging from about 3-5 to 7-5 kg. It is argued that Vulpavus was well adapted for climbing and was possibly arboreal, with locomotor behaviours comparable to those of the coatimundi (Nasua). Didymictis , on the other hand, was primarily terrestrial and probably incipiently cursorial. No modern taxon is similar to Didymictis in all aspects of the postcranial skeleton, but the Oriental civet ( Viverra ) is probably a reasonable modern analogue. Living members of the order Carnivora possess an array of postcranial specializations that enable particular taxa to exploit habitats ranging from marine, in the case of pinnipeds, to arboreal, exemplified by the prehensile-tailed kinkajou ( Potos ) and binturong ( Arctictis ). This morphological diversity and associated locomotor behaviours, probably derive from both of the early Tertiary carnivoran families, Miacidae and Viverravidae (Wortman and Matthew 1899; Flynn and Galiano 1982; Hunt and Tedford 1993), collectively termed miacoids. Although separate miacid and viverravid lineages extend back into the late Cretaceous (MacIntyre 1966; Fox and Youzwyshyn 1994), most of what has been known about miacoid postcrania is based on material no older than mid Eocene (Matthew 1909; Clark 1939; Springhorn 1980, 1982, 1985). Exceptions to this are a discussion of a fragmentary innominate and proximal femur belonging to the Palaeocene viverravid Protictis haydenianus (MacIntyre 1966), and very brief descriptions of Vassacyon promicrodon (Matthew 1915), and two species of Didymictis , D. altidens (Scott 1888 ; Matthew 1915) and D. protenus (Matthew 1901). More recently, however, and owing to fieldwork conducted over the past 15 years in the Willwood Formation of the Bighorn Basin, north-western Wyoming (Bown et a/. 1994), the amount of early Eocene miacoid postcranial material has increased significantly (Rose 1990; Heinrich 1995; Heinrich and Rose 1995). Two genera are particularly well represented in these new collections, the miacid Vulpavus and the viverravid Didymictis. Didymictis is known from both North America and Europe (Savage and Russell 1983). It first appears in the latest Paleocene (Clarkforkian North American Land Mammal Age, NALMA) and its temporal range extends through the early Eocene (Gingerich and Winkler 1985). Vulpavus on the other hand is an exclusively North American taxon known from both early (Wasatchian NALMA) and mid (Bridgerian NALMA) Eocene sediments (Gingerich 1983). It is the postcranial anatomy and locomotor behaviour of these two genera that is the focus of the present analysis. We describe their appendicular postcrania, comparing and contrasting their morphologies with one another and with an array of modern small- and medium-bodied carnivorans. MATERIALS Most of the fossil postcrania analysed and described in this study are from the early Eocene Willwood Formation, and are housed at the following institutions: the US Geological Survey IPalaeontology, Vol. 40, Part 2, 1997, pp. 279-305| © The Palaeontological Association 280 PALAEONTOLOGY, VOLUME 40 (USGS), now housed at the Department of Paleobiology, National Museum of Natural History, Smithsonian Institution; the American Museum of Natural History (AMNH), New York; and the University of Michigan (UM), Ann Arbor. This material was supplemented, however, by several North American mid Eocene miacids from the collections of the AMNH and the National Museum of Natural History (USNM), Washington. Extant carnivorans from the mammalogy departments of the AMNH and USNM as well as from the personal collections of the authors were used for comparative purposes. Among the 34 extant taxa analysed were representatives of each of the carnivoran families Mustelidae, Procyonidae, Canidae, Viverridae, Herpestidae and Felidae. Although Vulpavus and Didymictis are in need of taxonomic revision, the only species presently recognized in the Willwood Formation are V. australis, V. canavus and D. protenus, and we tentatively assign all of the early Eocene material described here to these three taxa. Vulpavus canavus may, on average, be slightly larger than V. australis , but otherwise there appear to be no specific differences in postcranial morphology. Several of the postcranial elements of early Eocene Vulpavus , however, are known only from poorly preserved and/or incomplete specimens and, therefore, in several instances we have figured bones of two other Wasatchian miacids, Miacis petilus (USGS 7161, distal tibia and fibula) and Uintacyon massetericus (USGS 21910, distal humerus), and two Bridgerian miacids, Miacis parvivorus (AMNH 1 1496, lunar) and a specimen attributed to Vulpavus sp. (USNM 362847, scapula, proximal tibia, and astragalus). Although this latter specimen lacks an associated cranium and dentition, the appendicular skeleton is well preserved and nearly complete and its overall similarities to other Bridgerian Vulpavus specimens leave little doubt that USNM 362847 belongs to this genus. Abbreviations of osteological features shown in Text-figures 1-8 are given in the Appendix. BODY MASS ESTIMATES The importance of reliable estimates of body mass for inferring locomotor behaviour and life history parameters of fossil taxa has been discussed at length (Damuth and MacFadden 1990, and references cited therein). Although these estimates have generally been calculated from regressions derived for dental dimensions (Gingerich et al. 1982; Legendre and Roth 1988; Van Valkenburgh 1990), it is intuitively obvious that a strong correlation exists between the size of a terrestrial mammal and the magnitude of mechanical loads that act on its limb bones. Body mass estimates for fossil taxa, therefore, have increasingly relied on extant mammal regressions of body mass on cross sectional parameters of various limb elements (Rulf et al. 1989; Anyonge 1993; Biknevicius et al. 1993; Runestad 1994; Heinrich and Rose 1995). Methods for obtaining cross sectional data used to derive the regressions employed here have been described in detail elsewhere (Heinrich 1995; Heinrich and Rose 1995). Briefly, cortical bone area and several measures of the distribution of cortical bone in cross section (i.e. second and polar moments of area) were collected at femoral midshaft and just below humeral midshaft for 24 extant caniform taxa (Table 1). Three methods were used to obtain these data: physical sectioning of bones, computer tomographic (CT) images, and biplanar X-rays. For the first of these, the cross section of a transversely sectioned bone was photographed and the endosteal and periosteal outlines digitized using a modified version of the computer program SLICE (Nagurka and Hayes 1980; Ruff and Hayes 1983) which calculates automatically the cross sectional parameters of interest. CT scans were digitized in the same way. Where bones could not be sectioned physically or subjected to CT scanning, the humerus and femur were modelled as hollow beams with a concentrically positioned medullary cavity, and biplanar X-rays were used to obtain medullary diameters (measured to the nearest 0T mm) in the anteroposterior and mediolateral planes. Section properties could then be estimated using standard geometrical formulae for an ellipse (Timoshenko and Gere 1972), a method which has been shown to provide accurate estimates of diaphyseal cross sectional properties (Runestad et al. 1993; Heinrich 1995). Least squares regressions of body mass on cortical area and polar moment of area using log- transformed species mean values for the 24 extant taxa analysed are given in Table 2. As a means HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 281 table 1. Extant carnivoran taxa for which cross sectional data of the humerus and femur were collected. N, sample size per species; Technique, method of obtaining cross sectional data. Taxon Common name N Technique Family Mustelidae Spilogale putorius Spotted skunk 10 X-ray Martes americana Pine marten 20 Physical section Mustela vison American mink 10 X-ray let onyx striatus Zorilla 5 X-ray Mephitis mephitis Striped skunk 9 X-ray Melogale personata Ferret badger 5 X-ray Galictis vittata Greater grison 5 X-ray Martes pennanti Fisher 20 Physical section Conepatus mesoleucus Hog-nosed skunk 10 X-ray Eira eira Tayra 10 X-ray Tax idea taxidea North American badger 11 X-ray Meles meles European badger 5 X-ray Gulo gido Wolverine 20 Physical section Family Procyonidae Bassariscus astutus Ringtail 10 X-ray Bassaricyon gabbi Olingo 5 X-ray Pot os flavus Kinkajou 5 X-ray Ailurus fulgens Fesser panda 6 X-ray Procyon lotor Raccoon 8 X-ray Family Canidae Fennecus zerda Fennec fox 6 CT scan Alopex lagopus Arctic fox 10 Physical section Urocyon cinereoargenteus Gray fox 10 CT scan Vulpes vulpes Red fox 10 Physical section Cerdocyon thous Crab-eating fox 10 X-ray Canis latrans Coyote 10 CT scan table 2. Regressions of log-transformed body mass on log-transformed cross sectional properties of the humerus and femur based on taxa given in Table 1 . r, correlation coefficient ; SE, standard error of regression ; % SEE, percentage standard error of estimate. Cross sectional property Slope Intercept r SE % SEE Humerus Cortical area 1-259 - 1 -270 0-980 0080 20-2 Polar moment of area 0-633 -0-987 0-986 0067 16-7 Femur Cortical area 1-326 - 1-308 0-968 0101 26-2 Polar moment of area 0-663 — 1-019 0-963 01 10 28-8 of assessing each equation’s ability to estimate accurately the dependent variable, the percentage standard error of estimate (percent SEE) was calculated for each regression where the percent SEE = antilog (2 + the standard error of the regression)— 100 (Smith 1984; Van Valkenburgh 1990). For both sectional properties the percentage SEE is less for the humeral than for the comparable 282 PALAEONTOLOGY, VOLUME 40 femoral regression (Table 2) suggesting that among carnivorans sectional properties of the humerus provide a better estimate of body mass than do those of the femur. Body mass estimates based on the four regressions given in Table 2 range from 3-4 kg to 6-6 kg (average 4-5 kg) for four specimens of Vulpavus (includes three specimens of V. canavus and one of V. australis) and from 3-9 kg to 8-3 kg (average 5-5 kg) for eight specimens of Didymictis protenus (Heinrich 1995). If only humeral regressions are considered, the range of body masses calculated for the eight specimens of D. protenus is 3-9 kg to 7-2 kg (average 5-0 kg), whilst the range for Vulpavus remains the same. Skeletal material of early Eocene Willwood Formation Vulpavus and Didymictis , therefore, suggests that these two fossil carnivorans were comparable in size to the living carnivorans Nasua nasua (coatimundi) and Vulpes vulpes (red fox) respectively. FORELIMB MORPHOLOGY Description and comparisons Scapula. Only fragmentary scapulae of Wasatchian Vulpavus and Didymictis are known, and descriptions here are limited to the glenoid region of that bone, the surface that articulates with the humeral head. In Vulpavus , the glenoid fossa is shallow and elliptical to pyriform-shaped, being wider posteriorly (i.e. at the axillary border) than anteriorly (Text-fig. 1a). In contrast, the glenoid text-fig. 1. Right scapulae of Vulpavus (A, USNM 362847) and Didymictis (b, USGS 5024, reversed) in lateral (left) and proximal (right) views. See Appendix for abbreviations. Scale bar represents 10 mm. fossa of Didymictis is rounder in outline (medial margin is expanded) and the supraglenoid tubercle (an attachment site for m. biceps brachii) extends well beyond the posteriormost aspect of the glenoid fossa. As a result of the supraglenoid morphology, the fossa of Didymictis is notably more concave anteroposteriorly than that of Vulpavus (Text-fig. 1b). Taylor (1974) noted an association among viverrids and herpestids of a deep glenoid with terrestrial adaptation, and a shallow glenoid fossa with arboreality. The suprascapular notch of Vulpavus is weakly developed so that scapular neck width (i.e. the shortest distance between axillary and anterior borders dorsal to the glenoid) is greater than the maximum length of the glenoid fossa (Text-fig. 1a). This contrasts with Didymictis where the suprascapular notch is deeper and neck width noticeably less than glenoid fossa length (Text-fig. 1b). Among modern carnivorans studied, neck width is always greater than glenoid length, but neck width tends to be greater in arboreal than in terrestrial taxa (Heinrich 1995). In Didymictis the base of the scapular spine is relatively farther from the glenoid margin than in Vulpavus , but in both the spine is closer to the axillary than to the anterior border (Text-fig. 1a-b). More complete scapulae that include the acromion process are known for several Bridgerian specimens of Vulpavus (e.g. AMNH 11498 and USNM 362847). In these animals the acromion HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 283 extends well beyond the glenoid fossa as in Procyon suggesting a well-developed m. acromiodeltoid and strong abduction (i.e. movement of the forelimb away from the midline of the body as opposed to adduction or movement towards the midline) capability. Wang (1993) also has described a relatively well developed clavicle (in AMNH 1 1498), a bone that is vestigial or absent in all modern Carnivora (Davis 1964). Unfortunately, the scapular spine is not well enough preserved in any specimen of Didymictis to discern the morphology of the acromion and no clavicle has yet been described. Humerus. Humeral head shape is similar for Vulpavus and Didymictis , but the size and orientation of the greater and lesser tuberosites differ markedly. In proximal view, the greater tuberosity of Vulpavus is narrow and forms a relatively wide angle with the sagittal plane of the diaphysis or humeral shaft, while the greater tuberosity of Didymictis is wider and oriented more anteroposteriorly (note superimposed lines in Text-fig. 2a, d). Unlike Vulpavus which is similar to extant arboreal (Text-fig. 3a) and most scansorial carnivorans in that neither tuberosity projects above the humeral head (Taylor 1974; Leach 1977; Laborde 1986, 1987), the greater tuberosity of the Didymictis humerus is well-developed, extending farther both anteriorly and proximally than in the miacid (Text-fig. 2b, d), and being quite similar in this respect to the morphology found in Can is (Text-fig. 3b) and Felis. The bicipital groove, for the tendon of m. biceps brachii, is better defined in Vulpavus than in Didymictis , a characteristic Vulpavus shares also with extant arboreal taxa such as Nandinia (Taylor 1974). Along the posteromedial aspect of the humerus of Vulpavus, just distal to the lesser tuberosity, is a relatively rugose insertion site probably for the muscles teres major and latissimus dorsi (Text-fig. 2b, d), large muscles involved in medial (i.e. internal) rotation of the shoulder joint and retraction of the forelimb (i.e. decreasing the distance between humeral shaft and axillary border of the scapula). The size and posteromedial flaring of this muscle attachment site gives the proximal diaphysis of Vulpavus a more markedly triangular cross sectional shape than the mediolaterally compressed humerus of Didymictis. Anteriorly, the deltopectoral crest of Vulpavus (on which insert important abductor, adductor, protractor and medial rotators of the forelimb) extends distally as a raised crest of bone that flares laterally just before ending abruptly at about midshaft (Text-fig. 2b, d). This morphology is not found among modern carnivorans but is very similar to that of the opossum Didelphis and the arboreal early Tertiary arctocyonid Chriacus (Rose 1987). In Didymictis the deltoid (i.e. lateral) margin of the deltopectoral crest is sharper and more distinct than the pectoral (i.e. medial) margin, and the deltopectoral crest is only slightly raised above the humeral shaft. Relatively wider proximally and occupying more of the anterior surface of the humerus than that of Vulpavus , the deltopectoral crest of Didymictis tapers distally and merges into the diaphysis well proximal to the midshaft (Text-fig. 2d), as occurs in extant cursorial carnivorans (Text-fig. 3b). The distal humerus of miacoids has a large entepicondylar foramen and is transversely broad with a well-developed medial epicondyle (better developed in miacids than in Didymictis ), the origin of m. pronator teres and the forearm and digital flexors (Text-fig. 2c, F). Large medial epicondyles are characteristic of carnivorans that climb and dig (Taylor 1974), are present but less well developed in Felis and Viverra , and are all but lost, along with the entepicondylar foramen, in cursorially specialized taxa like Vulpes and Cams (Text-fig. 3b). The lateral supinator crest of Vulpavus and Didymictis humeri are relatively wide and extend approximately one-half and one-third the length of the humerus respectively (Matthew 1901, 1915), providing a large attachment site for the flexor m. brachioradialis and the forearm and digital extensors. The size of the supinator crest in these miacoids is more similar to those of extant arboreal and scansorial taxa than to those of cursorial specialists (Text-fig. 3). The capitulum (for articulation with the radial head) is relatively wide and cylindrical in miacoids (Text-fig. 2c, f), but the trochlea (for articulation with the ulna) of at least some early Eocene miacids such as Uintacyon , and all of the known Bridgerian Vulpavus (Matthew 1909, figs 26, 36) differs from that of Didymictis in having a medial trochlear rim that extends only minimally beyond, and at a relatively shallow angle to the capitulum (note difference in angles formed by the 284 PALAEONTOLOGY. VOLUME 40 text-fig. 2. Right proximal and distal humerus. Proximal humeri of Vulpavus (a-b, USGS 25219) and Didymictis (d-e, USGS 5024) in proximal (a and d), anterior (b and e left) and medial (b and e right) views. Angles formed by greater tuberosity and sagittal plane of the humerus (i.e. superimposed lines in a and d) are approximately 60° for USGS 25219 and 40° for USGS 5024. Distal humeri of Uintacyon (c, USGS 21910, supinator crest reconstructed from Vulpavus USGS 16488) and Didymictis (f, USGS 27585) in anterior (left) and posterior (right) views. Note steeper medial trochlear rim relative to the long axis of the capitulum (i.e. superimposed lines) in the viverravid than the miacid. See Appendix for abbreviations. Scale bar represents c. 10 mm. superimposed lines in Text-fig. 2c, F). This is similar to the morphology found in arboreal carnivorans (Text-fig. 3a). Other early Eocene miacids, however, including Miacis petilus (Heinrich and Rose 1995), Vassacyon promicrodon and possibly Wasatchian Vulpavus , possess a trochlear rim morphology more similar to those of Didymictis and the scansorial carnivorans Martes and Nasua. All miacids are similar to the arborealists Nandinia and Potos in having a well delineated coronoid HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 285 text-fig. 3. Anterior (left) and medial (right) views of humeri belonging to the extant carnivorans Arctictis binturong (a, USNM 49642) and Can is lupus (b, USNM 324994). See Appendix for abbreviations. Scale bars represent 50 mm. fossa (Text-figs 2c, 3a) proximal to the trochlea proper, suggesting habitual use of highly flexed forelimb postures. The coronoid fossa is absent in Didymictis and modern carnivorans like Cams (Text-fig. 3b). The extremely shallow olecranon fossa of miacids is similar to that of the extant Polos , and is in striking contrast to the perforate supratrochlear foramen of Didymictis (Text-fig. 2f). The combination of perforate foramen and deeply grooved and angled (rather than proximodistally aligned trochlea as in canids; Text-fig. 3b) posterior trochlea in the viverravid is similar to the morphology found in badgers such as Mellivora , and indicates a significantly greater range of extension at the elbow and an enhanced stability of the humeroulnar articulation relative to Vulpavus. Between the medial epicondyle and the trochlear rim posteriorly, is a pit for attachment of the ulnar collateral ligament, a structure which anchors the semilunar notch of the ulna to the humerus (Evans and Christensen 1979). This pit is considerably larger and deeper in Didymictis than in miacids (Text-fig. 2c, f). Ulna. The ulnae of Vulpavus and Didymictis are mediolaterally compressed along their entire length with posterior diaphyseal borders that are slightly convex proximally (opposite the semilunar notch) and, at least in Didymictis , concave more distally (Text-fig. 4c, G). The olecranon process is nearly straight in both taxa but relatively longer in Didymictis than in Vulpavus (and most extant carnivorans), providing the viverravid with increased leverage for the forearm extensor, m. triceps. Bridgerian specimens of Vulpavus differ from those of the lower Eocene in that the olecranon process proximal to the semilunar notch is inclined anteriorly (Matthew 1915) as in the most 286 PALAEONTOLOGY, VOLUME 40 text-fig. 4. Right proximal and distal radius, proximal ulna, and carpal bones. Proximal radii of Vulpavus (a, USGS 5025) and Didymictis (e, USGS 21836, reversed) in proximal (top) and anterior (bottom) views. Distal radii of Vulpavus (b, USGS 25219) and Didymictis (f, USGS 25039) in posterior (top, forearm pronated) and distal (bottom) views. Ulnae of Vulpavus (c, USGS 25219) and Didymictis (G, USGS 5024) in anterior (left) and lateral (right) views. Scaphoid and lunar bones (d and h) in proximal (top) and distal (bottom) orientations, the dorsal margin of the bones is towards the top of the page. Scaphoids of Vulpavus (USGS 5025) and Didymictis (USGS 25039), lunars of Miacis (AMNH 1 1496) and Didymictis (USGS 25038). See Appendix for abbreviations. Scale bars represent 5 mm. arboreal of extant carnivores (Taylor 1974; Van Valkenburgh 1987). The proximalmost aspect of the olecranon process is higher medially than laterally in both miacoids, but in Didymictis this proximal projection is enhanced and a distinct groove for the tendon of m. triceps brachii is present (Text-fig. 4g). A well-developed m. triceps tendinal groove is found in extant carnivorans where rapid and complete extension of the forearm are important during locomotion. The anconeal process of Didymictis , and particularly the lateral margin, projects farther anteriorly than that of Vulpavus , producing a deeper, more concave semilunar notch than in the miacid (Text-fig. 4c, g). This smaller radius of curvature results in a more congruent articulation between ulna and the humeral trochlea of Didymictis throughout the range of flexion and extension. HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 287 More specifically, a strong contact is created between the lateral margin of the anconeal process and the posteriorly projecting medial trochlear margin which effectively locks the anconeal process in the trochlea during extension. By contrast, in Vulpavus and the arboreal Polos , the semilunar notch is less well defined and the lateral anconeal margin is distal and posterior to the medial margin (Text-fig. 4c) resulting in a smaller area of articular contact between ulna and the humeral trochlea during extension. The radial notch (i.e. articular surface for radial head) of both Vulpavus and Didymictis is relatively flat. Vulpavus differs, however, in having a long and narrow anterolaterally facing radial notch whilst that of Didymictis is nearly as wide proximodistally, as long anteroposteriorly and somewhat more anteriorly directed (Text-fig. 4c, G). A long and narrow muscle scar just distal to the coronoid process (Text-fig. 4c, G) of both miacoids is probably the insertion site of the forearm flexor m. brachialis. Distally, the anterior ulna of both Vulpavus and Didymictis flattens and widens, both sides bordered by sharp bony flanges. This large surface area implies a relatively well developed m. pronator quadratus in both taxa, a muscle that helps to pronate the forearm such that the palm of the forefoot is on the ground during locomotion. As in other miacids (Heinrich and Rose 1995), Vulpavus and Didymictis possess a flattened articular facet for the radius that is well separated from a second hemispherical facet at the end of the short, robust styloid process. The only obvious difference between the distal ulnae of Vulpavus and Didymictis is the presence in the miacid of a bony flange that projects posteriorly from the carpal articular facet of the styloid process. Radius. The radial head of Vulpavus is remarkably round with a small capitular eminence (Text-fig. 4a), and the proximal surface is oriented obliquely to the long axis of the radial shaft (Matthew 1915, fig. 28), with the posterolateral margin being notably higher than the anteromedial margin (forearm in pronation). In contrast, the proximal surface of the radial head of Didymictis is nearly perpendicular to the diaphysis, its outline is strongly elliptical, with the margin for articulation with the radial notch of the ulna being much less convex than that of Vulpavus , and the capitular eminence is much better developed (Text-fig. 4e). Analysis of radial head outline shape demonstrates similarities between Vulpavus and extant arboreal mammals capable of considerable supination on the one hand, and Didymictis and cursorial and fossorial forms on the other (MacLeod and Rose 1993). Proximally, the radial diaphysis of Vulpavus is relatively circular in cross section compared to the mediolaterally compressed diaphysis of Didymictis. The bicipital tuberosity is more prominent and located farther from the radial head in Vulpavus than in Didymictis , suggesting more powerful forearm flexion in the miacid. The distal radius of Vulpavus is wide and flat posteriorly and convex anteriorly, giving it a semilunar shape in cross section (Text-fig. 4b). In contrast, the distal diaphysis of Didymictis is almost triangular in cross section, with a considerably wider lateral (i.e. ulnar) margin than Vulpavus (Text-fig. 4f), and in this respect closely resembles radii of herpestids and Viverra. The facet for articulation with the ulna is somewhat larger and more concave in Didymictis than Vulpavus. Unlike Vulpavus , the radiocarpal surface of Didymictis possesses an expanded scaphoid articular surface that extends over the posteromedial lip as a convex facet (Text-fig. 4f). In modern carnivorans such as canids and felids, this expanded articular surface is even more prominent and functions to increase the range of flexion possible at the radiocarpal joint (Yalden 1970). Carpus. The carpus of early Eocene miacoids is poorly known, and descriptions and comparisons of carpal morphology are limited here to the scaphoid and lunar. Although fusion of these two bones along with the centrale into a single scapholunar bone is common to all living Carnivora (Flower 1871), among Wasatchian and Bridgerian miacids a fused scapholunar has been described only for Vassacyon promicrodon (Matthew 1915) and Vulpavus profectus (Matthew 1909), although fusion of the centrale and scaphoid has also been noted for Palaearctonyx meadi (Matthew 1909). The scaphoid and lunar are not fused in the single Vulpavus specimen known from the Willwood Formation, and Scott (1888) had previously noted that none of the three bones is fused in PALAEONTOLOGY, VOLUME 40 text-fig. 5. Middle (left) and ungual (right) phalanges of Vulpavus (a, USNM 362847 and USGS 5025 respectively) and Didymictis protenus (B, USGS 27585 reconstructed in part from AMNH 2855) in dorsal (top) and lateral (bottom) views, and Didymictis altidens (f, AMNH 14781 after Matthew 1915, text-fig. 19) in lateral view. Unguals of extant carnivorans Mcirtes pennanti (c, USNM 188226), Viverra zibetha (d, USNM 240208), and Canis latrans (E, USNM 49889). See Appendix for abbreviations. Scale bar represents 5 mm; c-F not drawn to scale. Didymictis. The sporadic occurrence of a fused scapholunar among miacoids led Flynn et al. (1988) to conclude that the scapholunar was acquired independently in various carnivoran lineages. 1. Scaphoid. The proximal scaphoid of Vulpavus (for articulation with the radius) is not uniformly convex, as in Didymictis , but relatively flatter dorsally than ventrally. The dorsalmost aspect possesses a slight lip (Text-fig. 4d) which would have come into contact with the radius during maximal extension at the radiocarpal joint. Laterally, the proximal scaphoid articular surface of Vulpavus is wide and convex but it narrows medially, pinching off at the base of the medially projecting scaphoid tubercle (Text-fig. 4d). In contrast, the proximal scaphoid of Didymictis is nearly rectangular in outline and lacks a lip along the dorsal margin (Text-fig. 4h). The prominent scaphoid tubercle extends more ventrally than laterally, and there is some slight expansion of the articular surface onto the base of the tubercle (Text-fig. 4h). This latter expansion is not as well developed as in canids and felids, where a distinct concave articular surface between tubercle and proximal scaphoid (Yalden 1970; Gonyea 1978) comes into contact with the posterior margin of the radiocarpal surface during hyperflexion at this joint (Yalden 1970), but the expanded articular surface found on the Didymictis scaphoid probably functioned in a similar manner. The lunar facet of the scaphoid is relatively larger in Vulpavus than in Didymictis , but in both it is flat and oriented distolaterally (Text-fig. 4d, h). Medial to the lunar facet is the centrale facet which in Vulpavus extends as a slightly concave articular surface from the dorsal to the ventral margins of the scaphoid. In Didymictis the centrale facet is divided by a small ridge into a larger, oval, ventral facet and a smaller, more elongate, dorsal facet (Text-fig. 4d, h). Medial to the ventral aspect of the centrale facet in both miacoids is the small, rounded trapezium facet (Text-fig. 4d, h). The trapezoid of Vulpavus probably articulated only with the centrale, while in Didymictis the trapezoid may also have articulated with the scaphoid (Text-fig. 4d, h). HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 289 2. Lunar. Whilst the proximal articular surface of the miacid lunar is relatively narrow along its entire dorsoventral curvature, it is further restricted to the medial margin dorsally by a non-articular shelf of bone (Text-fig. 4d). This non-articular shelf, much reduced in Didymictis (Text-fig. 4h). probably served as a ligamentous attachment site. In both miacoids the lateral margin of the lunar has two articular facets, a relatively narrow, fiat facet for the cuneiform (not well preserved for the available Didymictis specimen, LISGS 25038), and a more distally oriented, concave articular surface for the unciform. In miacids, the unciform facet is wider dorsally than ventrally (Text-fig. 4d). The lunar of miacoids has two additional articular surfaces, a large, distally oriented magnum facet and a small, more medially oriented facet for articulation with the centrale (Text-fig. 4d). In miacids the magnum facet is triangular, being wider ventrally than dorsally, and moderately concave except for its dorsalmost aspect which is nearly flat (Text-fig. 4d). This flat articular surface, set at an angle to the remainder of the magnum facet, is reminiscent of the morphology found m modern carnivorans in which a stop mechanism between magnum and lunar exists to prevent hyperextension at the midcarpal row (Yalden 1970). The magnum facet of Didymictis is rectangular rather than triangular, and more concave than that of miacids (Text-fig. 4h). No stop mechanism is apparent but this may be due to incomplete preservation of the bone. The centrale facet of the lunar is relatively larger and more nearly perpendicular to the magnum facet in Didymictis than in miacids (Text-fig. 4h). Metacarpcds and phalanges. Metacarpals and phalanges are poorly preserved for the miacoids of the Willwood Formation, preventing comparisons of their relative proportions both with one another and with other forelimb elements. In general, metacarpal (and metatarsal) diaphyses of Didymictis are dorsoventrally flatter than those of Vulpavus , and the proximal phalanges appear to differ only in that the distal articular surface is more deeply grooved in Didymictis than in Vulpavus. There are, however, several notable differences in the middle and ungual phalanges of these two fossil carnivorans. In addition to having a less prominent dorsoventral median ridge on the proximal articular surface, the middle phalanges of Didymictis differ from those of Vulpavus in having a more noticeable articular condyle asymmetry and a distinct flattening of the dorsal surface proximal to the articular head (Text-fig. 5a-b). This asymmetry and dorsal flattening of the middle phalanges of Didymictis resembles, superficially, that of the phalanges of felids and viverrids. carnivorans that possess retractile claws (Gonyea and Ashworth 1975). The middle phalanges of Didymictis differ from those of these modern carnivorans, however, in lacking the excavated lateral margin of the phalanx past which the ungual is retracted. The ungual or terminal phalanges of Vulpavus and Didymictis protenus are similar in being strongly compressed mediolaterally, in having a strongly curved dorsal margin (particularly near the tip of the bone), in possessing relatively well developed dorsal extensor and ventral flexor tubercles, and in lacking the dorsal fissure characteristics of extant fossorial mammals (Hildebrand 1985) and many creodonts (Denison 1938). They differ, however, in that the unguals of Didymictis protenus have a relatively wider proximal articular surface and the body of the phalanx is dorsoventrally narrower/shallower (Text-fig. 5a-b). Interestingly, the unguals of D. protenus are quite unlike those of the later early Eocene species Didymictis altidens (Text-fig. 5f), a morphological difference that does not appear to be expressed in other parts of the skeleton. The terminal phalanges of this younger taxon are relatively longer and less curved, and lack the well-developed flexor tubercle characteristic of D. protenus (Text-fig. 5b, f). These differences in ungual morphology may relate to fore- and hindlimb phalangeal comparisons (all unguals known for D. altidens are from the hindlimb while those attributed to D. protenus are indeterminate), but given that there is a number of unguals known for D. protenus , all with the same general morphology, this explanation for the morphological variation seems unlikely. Comparisons of lateral ungual outline shape between miacoids and modern taxa indicate that the unguals of Vulpavus are most similar to those of mammals classified as arboreal, while those of 290 PALAEONTOLOGY. VOLUME 40 D. protenus and D. altidens are generally similar to scansorial and terrestrial mammals respectively (MacLeod and Rose 1993). There is, however, considerable overlap between ungual morphologies and some types of locomotor behaviour, and although D. protenus shares many ungual attributes with scansorial taxa such as Martes (Text-fig. 5c), its unguals are not unlike those of the more strictly terrestrial and moderately cursorial Viverra (Text-fig. 5d). The unguals of D. altidens , on the other hand, resemble those of cursorial carnivorans like Can is (Text-fig. 5e) and, probably to an even greater extent, the terminal phalanges of semifossorial badgers like Meles (MacLeod and Rose 1993). Functional morphology of the forelimb Glenohumeral joint and brachium. The shoulder morphology of Vulpavus differs from that of Didvmictis in ways that indicate a greater range of motion at the glenohumeral joint and increased leverage of extrinsic shoulder musculature acting on the arm. Relatively greater mobility at this joint is inferred for Vulpavus from the shallow glenoid fossa and the low and angled greater and lesser tuberosities, both of which are indicative of habitually employed medial and lateral rotation at the shoulder. Reduced mobility at the glenohumeral joint of Didvmictis is suggested by the deep, rounded glenoid fossa, and the proximally projecting and anteroposteriorly aligned greater tuberosity. Morphology of the greater tuberosity in particular, indicates restricted abduction of the humeral head and at the same time increased leverage of nr. supraspinatus, a muscle that helps stabilize the shoulder joint during terrestrial locomotion (Taylor 1974; Jenkins and Weijs 1979; Larson and Stern 1989, 1992). The insertion sites for flexor, protractor and abductor-adductor musculature are more prominent and positioned farther distally in Vulpavus than in Didvmictis , indicating that the miacid generated relatively larger forces with many of the muscles that cross the shoulder joint than did the viverravid. Specifically, Vulpavus possesses a high deltopectoral crest that extends to midshaft, a character that has been equated with enhanced climbing abilities in small carnivorans (Taylor 1 974), and a prominent posteriorly projecting crest of bone for insertion of mm. latissimus dorsi and teres major. Insertion sites for these same muscles in Didvmictis are much less well defined and their more proximal position suggests that muscular force was compromised in favour of speed of contraction (Hildebrand 1988). Humeroulnar and radioulnar joints. The large medial epicondyle, moderately well developed supinator crest, wide trochlea and capitulum, and angled olecranon fossa of the distal humerus characteristic of both Vulpavus and Didvmictis are traits common to carnivorans that habitually climb or dig. In addition, however, Vulpavus possesses a distinct coronoid fossa, only moderately grooved posterior trochlea, and a shallow olecranon fossa, all of which are indicative of an emphasis on flexed rather than extended forelimb postures, implying that the animal was adapted for climbing. In contrast, the humeroulnar joint of Didvmictis , with its perforate supratrochlear foramen, deeply grooved trochlea, long olecranon process, and trochlear rim that extends well distal to, and at a relatively sharp angle to the capitulum, possesses characters which serve to maximize forearm extension and increase stability of the humeroulnar joint by "locking’ the semilunar notch in the trochlea. The humeroulnar joint of Didvmictis , therefore, is most similar in morphology to that of semifossorial carnivorans such as badgers which produce large forces across this joint. At the proximal radioulnar joint, the nearly round radial head of Vulpavus suggests that this animal was capable of substantial supination (perhaps up to 180°), whereas the ovoid radial head and anterolaterally facing radial notch in Didvmictis indicate significantly less motion at this joint in the viverravid. The relatively large capitular eminence in Didvmictis may also have restricted rotation of the radial head (Davis 1964), although an alternative function of this structure may be to stabilize the elbow in flexion, with the capitular eminence coming into contact with the medial HEINRICH AND ROSE: EOCENE MIACOIDCARNIVORANS 291 capitular wall and preventing lateral movement of the ulna on the trohlea. In either case the well- developed eminence of Didymictis is also indicative of increased stability in the elbow relative to Vulpavus. Radiocarpal joints and maims. Several characters of the proximal carpal row indicate that the radiocarpal joint of Didymictis was modified primarily for flexion and extension while no such specializations are found in the wrist of Vulpavus. The proximal scaphoid of the miacid has a larger radius of curvature than that of Didymictis , and a dorsal lip that probably served to limit rather than increase extension as the scaphoid came into contact with dorsal margin of the radiocarpal articular surface of the radius. Flexion at the radiocarpal joint of Didymictis was enhanced by expansion of the scaphoid articular surface on to the base of the scaphoid tubercle. This results in an articulation between the scaphoid and distal radius that is comparable to that found in extant carnivorans, but which is particularly well developed in cats and dogs where it allows for a range of flexion sufficient to prevent contact between the foot and opposite forelimb as the foot swings forward to initiate the next step in the gait cycle (Yalden 1970). The morphology of the middle and distal phalanges of Vulpavus (specifically the well-developed median ridge of the proximal articular surface and lack of asymmetry in the distal articular condyles of the middle phalanx, and the strong flexor tubercle and deep body of the unguals), indicates strong symmetrical flexion of the phalanges, probably an adaptation for digging the claws into tree-trunks and limbs during climbing. The unguals of D. protenus and D. a/tidens display two very different morphologies. D. protenus , the smaller and older of the two taxa, has relatively short, curved unguals that are more similar to those of Vulpavus than those of D. altidcns and suggest that D. protenus may have done some climbing. The longer and less curved unguals of D. altidcns more closely resemble those of strictly terrestrial extant carnivorans, particularly semifossorial taxa, suggesting that this species was a more proficient scratch digger than D. protenus. The dorsally flattened middle phalanx and slight asymmetry of the condyles, common to both D. protenus and D. altidcns (but not found in Vulpavus). suggest that these animals could partially retract the unguals into a position on the dorsum of the middle phalanx. They were not, however, capable of fully retracting the unguals alongside the diaphysis of the middle phalanx as occurs in modern felids and viverrids. Retractile claws are considered to be important for manipulating prey during killing (Gonyea and Ashworth 1975). Given that fully retractile claws are not found in either miacid or viverravid carnivorans, claw retractibility is unlikely to have been the primitive condition for Carnivora as argued by Flynn et al. ( 1988). HINDLIMB MORPHOLOGY Description and comparisons Innominate. The acetabulum or hip socket of both Vulpavus and Didymictis is buttressed anterodorsally, and a prominent tubercle for origin of the thigh flexor and leg extensor m. rectus femoris, lies just in front of the acetabulum (Text-fig. 6a -b). Among modern carnivorans this tubercle is better developed in scansorial than terrestrial taxa (Laborde 1986). The ventral margin of the ilium is quite wide in both miacoids, providing a large surface for the origin of a second important thigh flexor, m. lliacus. The lateral aspect of the iliac blade, however, appears wider and more concave in Vulpavus than in Didymictis (Text-fig. 6a-b), suggesting a relatively greater adductor muscle mass in the miacid. Posterior and dorsal to the acetabulum is the ischial spine, a partial origin for the gemelli muscles which abduct and laterally rotate the thigh. This spine is significantly more robust and situated farther from the acetabulum in Vulpavus than Didymictis. Among modern carnivorans the ischial spine is particularly well developed and positioned further posteriorly in the arboreal taxa Polos and Arctictis. In contrast, cursorial carnivorans such as Vulpes and Cains which have much less mobility at the hip joint (Jenkins and Camazine 1977), have poorly developed ischial spines. A greater distance between ischial spine and hip joint increases the mechanical advantage of the gemelli PALAEONTOLOGY, VOLUME 40 292 text-fig. 6. Left innominates of Vulpavus (a, USGS 16488. reversed) and Didymictis (b, reconstructed from USGS 21835 and 6087) in lateral views. See Appendix for abbreviations. Scale bar represents 10 mm. musculature, hence abduction and particularly lateral rotation of the thigh was probably stronger in Vulpavus than in Didymictis. The ischial tuberosity, preserved only for Vulpavus (Text-fig. 6a), is relatively broad and heavily scarred along its margin, indicating strong muscle attachments for the extensors of the thigh. Femur. The femoral head of Vulpavus is quite round, the articular surface extends well onto the femoral neck, and the femoral neck is relatively short (Text-fig. 7b), characters shared with modern arboreal (Text-fig. 7a) and scansorial carnivorans. In contrast, the femoral head of Didymictis , and terrestrial taxa generally, has a greater radius of curvature (i.e. forms a less complete sphere than that of Vulpavus). the articular surface is restricted to the margin of the head or is minimally expanded onto the femoral neck, and the femoral neck is elongate (Text-fig. 7e-f). The greater trochanter of Didymictis projects above the femoral head, and the ridge of bone joining the head and greater trochanter is much narrower anteroposteriorly (particularly notable just medial to the greater trochanter) than that of the miacid (Text-fig. 7b, f). A greater trochanter that projects above the femoral head is common to cursorial and saltatoriai mammals (Howell 1944), where it enables m. gluteus medius to act as a powerful extensor particularly in the later stages of retraction (Taylor 1976; Evans and Christensen 1979). Among modern carnivorans, the only taxa found to have the distinct anteroposterior narrowing of bone between femoral head and greater trochanter are canids, which also possess a less spherical femoral head and long femoral neck as in Didymictis. The lesser trochanter, for insertion of m. iliopsoas, projects more posteriorly than medially in Didymictis , canids, herpestids and terrestrial viverrids, while in Vulpavus and extant arboreal and scansorial taxa the lesser trochanter is oriented medially (Text-fig. 7b, f). Taylor (1976) has suggested that a more medially directed lesser trochanter provides for increased mobility and specifically enhanced lateral rotation at the hip. The proximal femoral diaphysis of early Eocene miacoids, and particularly Didymictis , is bowed or medially inflected (Text-fig. 7b, f) as in creodonts (Denison 1938; Gebo and Rose 1993), and is quite unlike the straight femoral diaphysis of most modern taxa (Text-fig. 7b, f). This morphology is made more prominent by an enlarged third trochanter, the insertion site of the superficial gluteal muscle. The superficial gluteal muscle is of HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 293 text-fig. 7. Left femur and proximal tibia. Femora of Pciradoxurus hermaphroditus (a, USNM 49868) and Viverra zibetha (e, USNM 256673, greater trochanter restored from the right side), arboreal and terrestrial extant carnivorans respectively. Proximal femora of Vulpavus (b. USGS 7143, reversed) and Didymictis (F, USGS 6087, reversed) in proximal (top), anterior (left), and posterior (right) views. Distal femora of I ulpavus (c, USGS 7143, reversed) and Didymictis (G, USGS 25040, reversed) in anterior (top) and distal (bottom) views. Proximal tibiae of Vulpavus (o, USNM 362847, reversed) and Didymictis (H, USGS 5024) in medial (top) and lateral (bottom) views. See Appendix for abbreviations. Scale bars represent 25 mm. variable size and function among modern carnivorans, acting as a flexor and medial rotator of the thigh in Ailttropoda (Davis 1 964). primarily an extensor of the thigh in Cam's ( Evans and Christensen 1979), and an abductor of the thigh in Fed is (Gilbert 1968). The position and orientation of the third trochanter of Vulpavus and Didymictis suggests that the superficial gluteal muscle acted primarily as a flexor and medial rotator much as in bears. Distally, the femur of miacoids is characterized by having medial and lateral condyles of similar width and a rugose medial epicondyle, for attachment of the medial collateral ligament. The distal femur of Didymictis from patellar trochlea to posteriormost aspect of the femoral condyles is deeper anteroposteriorly than that of Vulpavus , and the trochlea of Didymictis is more deeply grooved (Text-fig. 7c, o), closely resembling the morphology found in canids, felids, herpestids and I'ivcrra (Text-fig. 7e). The wide and relatively flat patellar trochlea of Vulpavus is similar to that of extant scansorial and arboreal taxa such as Bassariscits and Pciradoxurus (Text-fig. 7a), as well as the arboreal early Eocene arctocyonid Chriacus (Rose 1987). Tibia. The proximal tibiae of Vulpavus and Didymictis are similar in having a saddle-shaped medial :ondyle (convex anteroposteriorly and concave mediolaterally) that is higher than the nearly flat 294 PALAEONTOLOGY, VOLUME 40 text-fig. 8. All elements from left side. Distal tibiae of Miacis (A, USGS 7161) and Didymictis (g, USGS 27585) in anterior (left), posterior (right), and distal (bottom) views. Distal fibulae of Miacis (b, USGS 7161) and Didymictis (h, USGS 16472) in medial (left) and posterior (right) views. Astragali of Vulpavus (c, USNM 362847) and Didymictis (i, USGS 27585) in dorsal (left), ventral (right), and distal (bottom) views. Calcanei of Vulpavus (d, USGS 7143, partly reconstructed from a second Vulpavus specimen, USGS 25186) and Didymictis (j, USGS 27585) in dorsal (top) and distal (bottom) views. Naviculars of Vulpavus (e, USGS 5025) and Didymictis (k, AMNH 2855) in proximal (top) and distal (bottom) views. Cuboids of Vulpavus (f, USGS 5025, reversed and partially reconstructed from Vulpavus, USGS 25186) and Didymictis (l, AMNH 2855, reversed) in dorsal (left), medial (right), and distal (bottom) views. See Appendix for abbreviations. Scale bars represent 5 mm. lateral condyle, and in having a sharp ridge that extends distally from the posterior border of the medial condyle, probably separating the m. tibialis posterior laterally from the knee flexor m. popliteus medially. The tibial tuberosity of Didymictis is narrower and projects farther anteriorly than that of Vulpavus , and the m. tibialis anterior fossa lateral to the tuberosity, is more deeply HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 295 excavated in the viverrid (Text-fig. 7d, h). In both of these respects Didymictis is similar to modern cursorial carnivorans. On the anteromedial surface of the tibial shaft is a large raised tubercle (probably the insertion site of m. popliteus) that is better developed and more distally situated in Vulpavus than Didymictis (Text-fig. 7d, g), suggesting more powerful flexion and medial rotation of the crus in the miacid. The tibial diaphysis of both Vulpavus and Didymictis is compressed mediolaterally along most of its length. The distal tibia of miacoids bears a raised tubercle that runs from the lateral margin of the bone obliquely to the anterior surface. This interosseous tubercle (Text-fig. 8a, g) is more prominent in Didymictis than in Vulpavus , and was probably an attachment site for a strong interosseous membrane or syndesmosis between tibia and fibula. The tubercle is reduced in those modern carnivorans in which it can be discerned at all (e.g. Potos). On the posterior and medial aspect of the tibia is a second, well-defined tubercle, lateral to which passed the m. tibialis posterior tendon. In Vulpavus and other miacids (Text-fig. 8a, g), as in modern scansorial and arboreal carnivorans, this tibialis posterior tubercle angles anteriorly well proximal to the tibial malleolus, whilst in Didymictis and cursorially adapted modern taxa, the tubercle continues distally as a straight ridge of bone until it reaches or nearly reaches the distal margin of the malleolus. We interpret this difference in morphology as effectively positioning the m. tibialis posterior tendon to act primarily as an invertor of the foot in Vulpavus (and extant taxa that habitually climb) and as a plantarflexor of the foot in Didymictis (and terrestrial extant taxa). The anterior and distalmost aspect of the tibial malleolus of Didymictis projects laterally as a small malleolar tubercle for articulation with the cotylar fossa of the astragalus (Text-fig. 8g, i), a morphology not developed in miacids. It is found, however, among extant carnivorans such as canids in which the tibioastragalar articulation is restricted to flexion and extension. In all miacoids the distal tibia for articulation with the astragalar trochlea is set at an angle to the long axis of the tibia, rather than being nearly perpendicular to it as is more typical of modern carnivores. In Didymictis this sloping articular surface is divided into a small, horizontal, medial facet and a wider (albeit anteroposteriorly shorter) and strongly angled lateral facet, whereas in miacids these two facets are more equal in width and of similar slope (Text-fig. 8a, g). Dividing these medial and lateral facets is a ridge of bone (tibial crest of Jenkins and McClearn 1984) that articulates in the groove of the astragalar trochlea. This tibial crest is prominent in Didymictis but almost undetectable in miacids (Text-fig. 8a, g). Among living carnivorans, the tibial crest is best developed in cursors where it stabilizes the tibioastragalar joint, and restricts rotation to flexion and extension, while in arboreal taxa such as Nandinia the ridge is poorly developed allowing for some adduction and abduction at this joint in addition to flexion and extension (Taylor 1976). Fibula. Distal fibular morphology of Vulpavus and most other miacids differs from that of Didymictis in that the posterior aspect of the fibula has a shallow, medially oriented peroneal groove (rather than a deeper laterally oriented groove), and lacks an articular facet for the calcaneum (Text- fig. 8b, h). Proximal to the astragalar facet is a variably developed articulation for the tibia. This tibial facet (Text-fig. 8b, h) is present but small in some specimens of Vulpavus (e.g. USGS 7143) and Didymictis (USGS 16472, AMNH 2855), while in others it is absent (e.g. USGS 5025, Vulpavus). In another specimen of Didymictis (USGS 27585) the distal tibia and fibula are nearly fused and the articular facets obliterated. The combination of a relatively small distal tibiofibular articulation (where present at all) and large interosseous tubercle (Text-fig. 8a, g), suggests that a distal tibiofibular synovial joint was not as well developed in early Eocene miacoids as in living felids, viverrids, ursids, and most mustelids (Barnett and Napier 1953; Taylor 1976; pers. obs.). Instead, support at the tibiofibular joint of miacoids was probably maintained by a strong fibrous syndesmosis, a morphology considered by Barnett and Napier (1953) to be primitive for placental mammals. Tarsus. The miacoid tarsus, like that of extant carnivorans, includes seven bones: astragalus, calcaneum, navicular, cuboid and three cuneiforms - ecto, meso and entocuneiform. Of these the 296 PALAEONTOLOGY, VOLUME 40 astragalus and calcaneum have received considerably more attention than any of the other tarsal bones, and both have been used to address questions of functional morphology and phylogenetic relationships among carnivorans (Matthew 1909, 1915; Szalay 1977; Flynn and Galiano 1982; Gingerich 1983; Flynn et at. 1988; Wang 1993). In addition to the astragalus and calcaneum, we describe the navicular and cuboid. 1 . Astragalus. The astragalar trochlea of Vulpavus , for articulation with the distal tibia, differs from that of Didymictis in being less well grooved, in that the lateral aspect of the trochlea does not expand as far posteriorly (Text-fig. 8c, i), and in the medial and lateral trochlear crests having very different rather than comparable radii of curvature (the medial crest being smaller). In Vulpavus the trochlear articular surface expands onto the lateral aspect of the lateral trochlear crest (Text-fig. 8c, i), a morphology which probably relates to a greater range of abduction and possibly inversion of the foot during plantarflexion as discussed further below. Unlike modern carnivorans in which the astragalar foramen is oriented posteriorly or is absent (Wang 1993), the astragalar foramen of miacoids is large and more dorsally positioned, particularly in Vulpavus (Text-fig. 8c, i). Soft tissue structures passing through this foramen (e.g. nerves or vessels) may well have limited the range of motion possible between astragalus and tibia as suggested by Wang (1993), but the degree to which plantarflexion was restricted by this structure is difficult to determine given the remainder of the tibioastragalar joint morphology. Specifically, the medial aspect of the trochlear articular surface expands posteriorly well beyond the astragalar foramen in all miacoids (Text-fig. 8a, g) as noted by Szalay (1977) and others. This enabled the anteroposteriorly longer medial facet of the distal tibia (Text-fig. 8a, g) to maintain contact through a substantial range of rotation before the shorter, lateral facet of the distal tibia came into contact with the astragalar foramen. We estimate that Vulpavus was capable of rotating the astragalus to a position of about 1 15° from the long axis of the tibia, while the angle between tibial diaphysis and the long axis of the astragalus probably exceeded 140° in Didymictis during maximum plantarflexion. On the medial aspect of the astragalus of Didymictis is a distinct, concave facet (the cotylar fossa) for articulation with the lateral aspect of the tibial malleolus (Text-fig. 8g, t). This fossa, not present in miacids, is associated with a short, narrow, and dorsally directed groove occupied by the posterior margin of the tibial malleolus when the two bones are articulated. This combination of cotylar fossa and groove has also been noted for the early Eocene arctocyonid Anacodon (Rose 1990) and at least some species of the mesonychid Pachyaena (O’Leary and Rose 1995), and a well- developed malleolus-cotylar fossa articulation characterizes many extant cursorial mammals, including canids. The sustentacular and ectal facets on the ventral surface of the astragalus articulate with the sustentaculum and posterior calcaneal facet of the calcaneum respectively, collectively forming the subtalar joints. The sustentacular facet of Vulpavus is more convex than that of Didymictis (which is nearly flat) while the ectal facet of miacids is strongly concave and helical in morphology (Text- fig. 8c), the posterior aspect facing more laterad and the anterior aspect distoventrad. In contrast, the ectal facet of Didymictis has a greater radius of curvature and lacks the helical orientation (Text- fig. 8i). The ectal and sustentacular facets are relatively farther apart in Vulpavus than in Didymictis , and the sustentacular facet of miacids is more distinctly anterior, there being little overlap between the posterior aspect of the sustentacular facet and the anterior aspect of the ectal facet. Sustentacular and ectal facet morphology of Vulpavus is similar to that found in the extant carnivoran Potos where it enhances inversion and eversion at the subtalar joints (Jenkins and McClearn 1984). Posterior and lateral to the ectal facet is the groove for the tendons of the plantarflexor mm. flexor hallucis longus and flexor digitorum longus. This groove is conspicuously deeper on the astragalus of Vulpavus than on that of Didymictis (Text-fig. 8c, i), but in both taxa it is oriented at an angle oblique to, rather than aligned with the long axis of the astragalar trochlea as occurs in extant canids (Wang 1993). The astragalar head of Vulpavus is flattened dorsoventrally and smoothly convex, with the articular surface for the navicular expanding farther onto the lateral aspect of the head than in HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 297 Didymictis (Text-fig. 8c). The astragalar head of Didymictis is rotated so that its long axis is more closely aligned to the parasagittal plane (Text-fig. 81), a character common to modern cursorial carnivorans. The astragalar head of Didymictis also possesses a distinct articular surface, absent or very reduced in Vulpavus, for contact with the calcaneum, as is also found in canids. 2. Navicular. The navicular of Vulpavus is notably wider than that of Didymictis , in large part due to the presence of a tubercle projecting from the dorsal and medial corner of the bone (Text-fig. 8e) probably for the insertion of m. tibialis posterior. A second tubercle is present ventrally, probably for insertion of the calcaneonavicular (spring) ligament. This latter tubercle is quite round in Vulpavus but anteroposteriorly elongate in Didymictis , extending anteriorly beyond the articular surface for the cuneiforms. Along the lateral margin of the navicular is a slightly concave articular surface for the cuboid. This facet is longer anteroposteriorly and more rectangular in Didymictis than in Vulpavus , resulting in a strong contact with very little movement between navicular and cuboid in the viverravid. The same facet in Vulpavus is helical in shape, the dorsal part facing directly laterally and the ventral part oriented somewhat more posterolaterally, and the articular surface narrows considerably between the dorsal and ventral aspects of the facet. Articulation of the cuboid and navicular indicates considerably more motion between these two bones in Vulpavus than was possible in Didymictis. The distal aspect of the navicular has three well-defined articular facets for the cuneiforms, the lateral or ectocuneiform facet being the largest in both miacoids (Text-fig. 8e, k). The cuneiform facets of Didymictis tend to be wider dorsally than ventrally, whilst in Vulpavus the same facets are more nearly square. The ento- and mesocuneiform facets are slightly convex dorsoventrally in both Vulpavus and Didymictis but unlike the convex curvature of the lateral cuneiform facet of Vulpavus , the ectocuneiform of Didymictis is nearly flat. 3. Calcaneum. Like the astragalus and navicular, the calcaneum of Vulpavus differs from that of Didymictis in several notable ways. The sustentaculum is relatively larger, more dorsally oriented, and located farther from the posterior calcaneal facet than that of Didymictis (Text-fig. 8d, j). Unlike Vulpavus , in which the posterior calcaneal articular surface comes into contact only with the ectal facet of the astragalus, in Didymictis the posterior calcaneal articular surface is divided by a distinct ridge into a medially oriented ectal facet and a dorsally oriented fibular facet (Text-fig. 8j). Just posterior to the fibular facet is a small pit where the posterior and distalmost aspect of the fibula contacts the calcaneum, providing the ultimate limiting factor in the range of plantarfl exion possible at the tibioastragalar joint of Didymictis , as noted by Hunt and Tedford (1993). The calcaneum of Didymictis , unlike that of Vulpavus , is elongated both proximal and distal to the subtalar joints (Text-fig. 8d, j), a morphology characteristic of cursorial and saltatorial mammals generally (Howell 1944; Hildebrand 1988). The peroneal tubercle of Didymictis , like that of canids, is small and situated distally (i.e. just lateral to the cuboid facet), and is in contrast to the much better developed and more proximally positioned peroneal tubercle of Vulpavus (Text-fig. 8d, j). This latter morphology is found among modern carnivorans well adapted for climbing, such as Potos , Nasua and Nandina , where the peroneal musculature, and particularly m. peroneus longus, functions to evert and abduct as well as plantarllex the foot. In dogs, the peroneal musculature acts primarily to plantarflex the pes (Evans and Christensen 1979). The cuboid facet of the calcaneum is at an acute angle to the long axis of the calcaneum in both Vulpavus and Didymictis but this angle is more acute and the cuboid facet flatter (rather than concave) in the viverravid than in the miacid (Text-fig. 8d, j). In both of these respects the calcaneum of Didymictis is more similar to those of herpestids and canids while the calcaneum of Vulpavus closely resembles that of some living scansorial and arboreal carnivorans. The calcaneum of Vulpavus also differs from that of Didymictis in (1) lacking a well-developed articular surface for the astragalar head along the dorsomedial aspect of the cuboid facet (Text-fig. 8d, j), (2) having a less 298 PALAEONTOLOGY, VOLUME 40 well defined groove for the flexor hallucis longus tendon along the ventral aspect of the sustentaculum tali, and (3) having a smaller plantar tubercle positioned at the margin of the cuboid facet rather than well proximal to it. Szalay (1977) argued that a plantar tubercle set well back from the cuboid facet, as in Didymictis , is a synapomorphic character of Creodonta and early carnivorans, but in fact the morphology cited by Szalay (1977) does not even characterize viverravids (pers. obs.) let alone miacoids. 4. Cuboid. Whereas the cuboid of Didymictis is relatively reactangular, that of Vidpavus is distinctly wider proximally than distally owing to a more laterally expanded proximal facet for articulation with the calcaneum (Text-fig. 8f, l). This, along with its more uniform convex morphology, allowed for a substantially greater degree of abduction at the transverse tarsal joint in Vidpavus than in Didymictis. On the medial aspect of the cuboid are facets for the astragalar head, navicular and ectocuneiform (Text-fig. 8f, l). The flat ectocuneiform facet of Didymictis is clearly demarcated from the navicular facet by an approximately 30° change in orientation. In Vidpavus the ectocuneiform facet is saddle-shaped (i.e. slightly concave proximodistally and convex dorso- ventrally) and the change in orientation between navicular and ectocuneiform facets is closer to 45°. The distal cuboid of miacoids articulated with the fourth and fifth metatarsals with the facet for the fourth being much larger than that for the fifth (Text-fig. 8f, l). In Didymictis both of these facets face distally whereas in Vidpavus the facet for the smaller fifth metatarsal faces more laterally than distally, indicating that the fifth digit of Vulpavus was capable of being abducted considerably farther than that of Didymictis. Ventrolaterally, and oriented more or less perpendicular to the long axis of the cuboid, is a well-developed tubercle, probably for insertion of the long plantar ligament (Text-fig. 8f, l). In Didymictis this tubercle is well separated from both the calcaneal and metatarsal facets, but in Vulpavus the tubercle is expanded laterally and nearly continuous with the lateral expansion of the proximal calcaneal facet. Functional interpretation of the hindlimb Hip and knee joints. Innominate and proximal femoral morphology indicate that the hip joints of Vulpavus and Didymictis were heavily muscled. The well-developed anterior iliac tubercle, wide ventral ilium, and distally positioned and laterally oriented third trochanter, suggest powerful flexion of the thigh in both taxa. Several characters indicate that Vulpavus , unlike Didymictis , regularly employed abducted and laterally rotated hip postures, as do extant scansorial taxa such as Procyon (Jenkins and Camazine 1977) as well as arboreal taxa. These include a well-developed and posteriorly positioned ischial spine, a more spherical femoral head, expansion of the femoral head articular surface onto the femoral neck, and a medially projecting lesser trochanter. In contrast, the smaller, more craniad ischial spine, reduced ilium, posteriorly directed lesser trochanter, and less spherical femoral head with a more restricted articular surface, indicate that the hip joint of Didymictis had a more limited range of motion and was probably involved in more strictly parasagittal gaits. The knee joint of Didymictis is characterized by anteroposteriorly deep femoral condyles, deeply grooved patellar trochlea, deep m. tibialis anterior fossa, and a mediolaterally compressed and anteriorly projecting tibial tuberosity. Although not as well developed as in canids, these characters are indicative of an emphasis on rapid flexion and extension of the knee. The knee morphology of Vulpavus on the other hand, with its mediolaterally wide and anteroposteriorly narrow femoral condyles, shallowly grooved patellar trochlea, and well-developed m. popliteus insertion site, is indicative of an emphasis on powerful flexion and more plantigrade and ambulatory locomotion (Ginsburg 1961). Ankle , subtalar and transverse tarsal joints. The pes of Vidpavus was adapted for mobility, specifically abduction-adduction and inversion-eversion, while the ankle and intertarsal joints of Didymictis suggest that motion was largely restricted to flexion and extension. The shallow trochlea HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 299 (and related flat articular surface of the distal tibia) and unequal curvatures of the medial and lateral trochlear crests of the Vulpavus astragalus indicate simultaneous inversion and abduction at the tibioastragalar joint during plantarflexion, as described for Potos by Jenkins and McClearn (1984). The range of flexion and extension possible in the ankle of Didymictis was greater than that of Vidpavus , characters indicative of an emphasis on plantarflexion at the tibioastragalar joint in the viverravid including (1) a more distally positioned astragalar foramen, (2) the increased length of the astragalar trochlea resulting from elongation of the trochlea laterally, (3) the presence of small, comparably developed radii of curvature of the medial and lateral trochlear crests, and (4) the more salient astragalar trochlear groove, associated with a well-developed tibial crest. Anterior translation or sliding of the astragalus on the calcaneum produced inversion at the subtalar joint of Vulpavus as the helically shaped ectal facet rotated from the anterolateral to posteromedially oriented aspects of the posterior calcaneal facet. In Potos this motion accompanies plantarflexion and, along with the abduction and inversion at the tibioastragalar joint, enables this animal to reverse its hindfoot completely when hanging from branches or descending vertical supports (Jenkins and McClearn 1984). Although there was substantial capacity for hindfoot inversion and abduction in early Eocene Vulpavus , it was probably incapable of complete hindfoot reversal for two reasons: (1) the restricted range of plantarflexion at the tibioastragalar joint due to the dorsal position of the astragalar foramen (the trochlea of Potos forms an almost 180° arc), and (2) the relatively short posterior calcaneal facet (the posterior aspect of this facet is notably longer in Potos). This latter character would have functioned to limit the extent of posterior translation and subsequent conjunct rotation in Vulpavus. In Didymictis the large fibular-calcaneal articulation prevented the ectal facet from articulating with the dorsally oriented and more anterior aspect of the posterior calcaneal surface, thereby reducing, if not effectively eliminating, inversion at the subtalar joint. Motion of the astragalus on the calcaneum, therefore, was limited to fore-and- aft translation on the medially facing ectal facet. The transverse tarsal joint of Vulpavus was also capable of a greater degree of inversion-eversion and abduction-adduction than that of Didymictis. Szalay and Decker (1974) argued that a lateral and dorsally expanded articular surface on a dorsoventrally flattened astragalar head as is characteristic of Vulpavus is indicative of habitual eversion, while the well-developed peroneal tubercle of this animal indicates enhanced eversion and abduction by increasing the mechanical advantage of the peroneal musculature. Conversely, the dorsoventrally rotated astragalar head of Didymictis reflects an emphasis on flexion and extension at the astragalonavicular articulation, and the similarity of peroneal tubercle morphology among Didymictis , herpestids, and canids suggests that the peroneal musculature in the viverravid may have functioned predominantly as a plantarflexor. Inversion at the transverse tarsal joint of Didymictis , as in the subtalar joints, also appears to have been restricted. Transverse tarsal inversion requires that the calcaneum, cuboid and navicular rotate as a unit about the astragalar head (Jenkins and McClearn 1984), an action limited in Didymictis by the combination of the cuboid-calcaneum articular morphology, and an interlocking tarsal organization, involving the astragalar head, calcaneum and cuboid, similar to that described for the hyaenodontid Gazinocyon (Polly 1996). Increased mobility at the transverse tarsal joint of Vulpavus is indicated by the mediolaterally wider and more concave facet of the proximal navicular, lack of a well-developed astragalar head-calcaneum articulation, and the more concave and better defined articulation between the calcaneum and cuboid. Although Hildebrand (1988, p. 478) stated that tarsal bones lengthen only in jumping mammals (although see the cheetah calcaneum; Hildebrand 1988, text-fig. 24-15), the clearly elongate calcaneum of Didymictis relative to that of Vulpavus is similar to the differences found between the calcanei of the cursorial Alopex and the arboreal Potos. The elongate tuber calcaneum of extant cursors increases the lever arm of the main plantarflexor musculature (i.e. mm. gastrocnemius, plantaris and soleus), indicating selection for power over speed of contraction at the tibioastragalar joint. This is contrary to what is more often the case in mammals adapted for speed: maximization of the velocity of rotation about distal joints. There are at least two reasons why cursors may 300 PALAEONTOLOGY, VOLUME 40 increase the lever arm of the main plantarflexors of the ankle: (1) to increase the force generated by the relatively reduced plantarflexor muscle mass associated with weight reduction of distal limb segments; and (2) to increase the mechanical advantage of these muscles during a prolonged digitigrade stance. This latter possibility raises the question of whether Didymictis was digitigrade, subdigitigrade or plantigrade. In a resting position, living digitigrade canids form an angle of about 150° between tibia and long axis of the tarsometatarsus (Wang 1993). Although Didymictis was probably incapable of creating an angle between tibia and tarsometatarsus of more than 150° during maximal plantarflexion, the significant increase in the amount of plantarflexion possible relative to miacids and the substantial elongation of the tuber calcaneum strongly indicate that this animal employed at least a semidigitigrade stance, as proposed by Matthew (1909). SUMMARY The appendicular skeleton of Wasatchian Vulpavus indicates that this animal possessed a considerable range of motion at most joints of the fore- and hindlimb, a degree of mobility that is similar to that found in modern carnivorans adapted for exploiting arboreal habitats (Text-fig. 9). Drawing by Jay H. Matternes © 1989 text-fig. 9. Reconstruction of the Bridgerian NALMA (middle Eocene) miacid Vulpavus ovatus. The postcranial skeleton of Didymictis , on the other hand, is suggestive of more restricted parasagittal motion (particularly in the hindlimb) and increased joint stability, as characterizes extant carnivorans adapted primarily for speed. There are, however, several characters in the forelimb that imply that Didymictis also employed substantial digging in its behavioural repertoire. The forelimb of Vulpavus is characterized by: a shallow glenoid fossa, low greater tuberosity, narrow and laterally flared deltopectoral crest, large supinator crest, large projecting medial epicondyle, a medial trochlear rim extending only minimally beyond the capitulum, shallow olecranon fossa, wide semilunar notch with a poorly defined lateral wall, flat and laterally facing radial notch, wide anterior ulna distally, round radial head, strong extensor tubercles and relatively shallow articular surface on the distal radius, and deep, laterally compressed, dorsally curved ungual phalanges with well-developed flexor and extensor tubercles. These characters imply the capability for powerful protraction-retraction, abduction-adduction, and medial-lateral rotation HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 301 about the glenohumeral joint and flexion-extension about the humeroulnar and radiocarpal joints, an extreme range of pronation-supination at the proximal and distal radioulnar joints, and strong, sharp claws, all of which are essential for climbing. Similarly, characters of the Vulpavus hindlimb are suggestive of considerable rotation at the hip joint and extensive abduction-adduction and inversion-eversion at the tibioastragalar and subtalar joints. These characters include: a large ischial spine, rounded femoral head with posteriorly expanded articular surface, medially directed lesser trochanter, relatively wide and shallow patellar trochlea, flat and inclined distal tibial articular surface, proximal and anteriorly oriented malleolar tubercle, medial and lateral astragalar trochlear crests with differing radii of curvature, laterally and dorsally expanded articular surface of the astragalar head, absence of a fibular-calcaneal articulation, large and laterally projecting peroneal tubercle, and helical morphology of the articulation between ectal and posterior calcaneal facets. Behavioural interpretations based on the forelimb and hindlimb of Didymictis , unlike those for Vulpavus , are not completely congruous. The forelimb possesses a deeper, more rounded glenoid fossa, increased articular congruence at the glenohumeral joint, a proximally projecting greater tuberosity, reduced deltopectoral crest, relatively well developed supinator crest (although less well developed than in Vulpavus ), wide medial epicondyle, deep and perforate olecranon fossa, steeply inclined humeral trochlear rim, large olecranon process having a well-defined triceps tendinal groove proximally, a deep semilunar notch, anterolaterally facing radioulnar notch and very oval radial head with prominent capitular eminence, a very concave distal radial articular surface expanded over the posterior margin as a convex articular surface for the scaphoid, and a scaphoid and lunar with small radii of curvature along their proximal articular surfaces. Relative to Vulpavus , all of these are indicative of reduced mobility at the glenohumeral, elbow and radiocarpal joints, where rotation and supination are sacrificed for stability during powerful flexion and extension. Although most of these characters resemble those found in extant cursors, the morphology of the distal humerus and the outline shape of the unguals of D. altidens are decidedly more similar to those of modern semifossorial taxa. The hindlimb of Didymictis is indicative of incipient cursoriality. The reduced ischial spine and ilium, less spherical femoral head with reduced articular surface, posteriorly directed lesser trochanter, and narrow ridge of bone between head and greater trochanter, all suggest reduced capacity for employing rotated and abducted hip postures. Restriction of hindlimb motion to a parasagittal gait is further enhanced by the narrow and deep patellar trochlea, more vertically oriented malleolar tubercle, medially projecting tibial malleolus, better defined tibial crest on the distal tibia articular surface, deep and posterolaterally facing peroneal groove, dorsoventrally oriented astragalar head, well-developed calcaneo-fibular articulation, reduction of the peroneal tubercle, and elongate calcaneum. All of these characters suggest an emphasis on flexion and extension at the expense of eversion-inversion and abduction-adduction. It seems likely, therefore, that Didymictis was a relatively specialized terrestrial carnivore capable of hunting either with speed or by pursuing its quarry by means of digging. Acknowledgements. We thank Drs T. Bown, R. Emry, P. Gingerich, and R. Tedford for granting access to specimens, and Drs B. Van Valkenburgh and Wang Xiaoming for helpful discussions on the functional morphology of early carnivorans. The manuscript was improved by comments of two anonymous reviewers. Dr R. Chapman kindly provided an IBM compatible version of the computer program SLICE. We are indebted to E. Kasmer for illustrating all of the fossil material and J. Matternes for allowing us to use his reconstruction of Vulpavus ovatus. This study derives in part from doctoral work conducted in the Department of Cell Biology and Anatomy, The Johns Hopkins University School of Medicine (Heinrich 1995) and completed with the support of the Department of Anatomical Sciences, SUNY at Stony Brook, NY. Field work and analysis has been supported by NSF grants BSR-8215099, 8500732, 8918755 and IBN-9419776. REFERENCES anyonge, w. 1993. Body mass in large extant and extinct carnivores. Journal of Zoology , 231, 339-350. 302 PALAEONTOLOGY, VOLUME 40 barnett, c. h. and napier, J. r. 1953. The rotatory mobility of the fibula in eutherian mammals. Journal of Anatomy, 87, 1 1—21. biknevicius, a. r., McFarlane, D. a. and MacPhee, r, d. e. 1993. Body size of Amblyrhiza inundata (Rodentia: Caviomorpha), an extinct megafaunal rodent from the Anguilla Bank, West Indies: estimates and implications. American Museum Novitates , 3079, 1-25. bown, t. m., rose, k. D., simons, E. l. and wing, s. l. 1994. Distribution and stratigraphic correlation of upper Paleocene and lower Eocene fossil mammal and plant localities of the Fort Union, Willwood, and Tatman Formations, southern Bighorn Basin, Wyoming. Professional Paper of the US Geological Survey. 1540, 1-103. Clark, j. 1939. Miacis gracilis, a new carnivore from the Uinta Eocene. Annals of the Carnegie Museum, 27, 349-370. damuth, j. and MacFADDEN, b. j. 1990. Introduction: body size and its estimation. 1-10. In damuth, j. and MacFADDEN, b. j. (eds). Body size in mammalian paleobiology : estimation and biological implications. Cambridge University Press, Cambridge, 397 pp. davis, d. d. 1964. The giant panda: a morphological study of evolutionary mechanisms. Fieldiana, Zoology Memoir, 3, 1-339. denison, R. h. 1938. The broad skulled Pseudocreodi. Annals of the New York Academy of Science, 37, 163-256. Evans, h. E. and Christensen, c. c. 1979. Miller s anatomy of the dog. W. B. Saunders, Philadelphia, 1181 pp. flower, w. h. 1871. On the composition of the carpus of the dog. Journal of Anatomy, 6, 62-664. flynn, J. J. and galiano, H. 1982. Phytogeny of early Tertiary Carnivora, with a description of a new species of Protictis from the middle Eocene of northwestern Wyoming. American Museum Novitates, 2725, 1-64. -neff, N. a. and tedford, r. h. 1988. Phylogeny of the Carnivora. 73-115. In benton, m. j. (ed.). The phytogeny and classification of the tetrapods. Volume 2, mammals. Clarendon Press, Oxford, 326 pp. fox, r. c. and youzwyshyn, g. p. 1994. New primitive carnivorans (Mammalia) from the Paleocene of western Canada, and their bearing on relationships of the order. Journal of Vertebrate Paleontology , 14, 382^404. gebo, d. l. and rose, k. d. 1993. Skeletal morphology and locomotor adaptation in Prolimnocyon atavus, an early Eocene hyaenodontid creodont. Journal of Vertebrate Paleontology, 13, 125-144. gilbert, s. G. 1968. Pictorial anatomy of the cat. University of Washington Press, Seattle, 120 pp. gingerich, p. d. 1983. Systematics of early Eocene Miacidae (Mammalia, Carnivora) in the Clark’s Fork Basin, Wyoming. Contributions from the Museum of Paleontology, University of Michigan , 26, 197-225. — smith, b. h. and rosenberg, k. 1982. Allometric scaling in the dentition of primates and predictions of body weight from tooth size in fossils. American Journal of Physical Anthropology, 58, 81-100. and winkler, d. a. 1985. Systematics of Paleocene Viverravidae (Mammalia, Carnivora) in the Bighorn Basin and Clark's Fork Basin, Wyoming. Contributions from the Museum of Paleontology, University of Michigan, 27, 87-128. ginsburg, l. 1961. Plantigrade et digitigrade chez les carnivores fissipedes. Mammalia, 25, 1-21. gonyea, w. j. 1978. Functional implications of felid forelimb anatomy. Acta Anatomica, 102, 111-121. and ashworth, r. 1975. The form and function of retractile claws in the Felidae and other representative carnivorans. Journal of Morphology, 145, 229-238. heinrich, R. E. 1995. Functional morphology and body size of early Tertiary Miacoidea (Mammalia, Carnivora). Unpublished Ph.D. Dissertation, Department of Cell Biology and Anatomy, Johns Hopkins School of Medicine, Baltimore, 241 pp. — and rose, k. d. 1995. Partial skeleton of the primitive carnivoran Miacis petilus from the early Eocene of Wyoming. Journal of Mammalogy, 76, 148-162. hildebrand, m. 1985. Digging of quadrupeds. 89-109. In Hildebrand, m., bramble, m., liem, k. and wake, d. (eds). Functional vertebrate morphology. Belknap Press, Cambridge, 430 pp. 1988. Analysis of vertebrate structure. John Wiley and Sons Inc., New York, 701 pp. howell, A. b. 1944. Speed in animals. Hafner Press, New York, 270 pp. hunt, R. m., jr and tedford, r. h. 1993. Phylogenetic implications within the aeluroid Carnivora and implications of their temporal and geographic distribution. 53-73. In szalay, f. s., novacek, m. j. and McKenna, m. c. (eds). Mammal phylogeny : placentals. Springer-Verlag, New York, 321 pp. jenkins, f. a., jr and camazine, s. m. 1977. Hip structure and locomotion in ambulatory and cursorial carnivores. Journal of Zoology, 181, 351-370. — and McClearn, d. 1984. Mechanisms of hind foot reversal in climbing mammals. Journal of Morphology, 182, 197-219. and weijs, w. a. 1979. The functional anatomy of the shoulder in the Virginia opossum (Didelphis virginianus). Journal of Zoology, 188, 379-410. HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 303 laborde, c. 1986. Description de la locomotion arboricole de Cryptoprocta ferox (Carnivore Viverride nralgache). Mammalia , 50, 369-378. 1987. Caracteres d’adaptation des membres au mode de via arboricole chez Cryptoprocta ferox par comparison avec d’autres Carnivores viverrides. Annales des Sciences Naturelles , Zoologie , Paris , 8, 25-39. larson, s. G. and stern, j. t., Jr 1989. Role of supraspinatus in the quadrupedal locomotion of vervets (Cercopithecus aethiops): implications for interpretation of humeral morphology. American Journal of Physical Anthropology , 79, 369-377. 1992. Further evidence for the role of supraspinatus in quadrupedal monkeys. American Journal of Physical Anthropology , 87, 359-364. leach, d. 1977. The forelimb musculature of marten ( Martes americana Turton) and fisher ( Martes pennanti Erxleben). Canadian Journal of Zoology , 55, 31-41. legendre, s. and roth, c. 1988. Correlation of carnassial tooth size and body weight in recent carnivores (Mammalia). Historical Biology, 1, 85-98. MacIntyre, g. t. 1966. The Miacidae (Mammalia: Carnivora). Bulletin of the American Museum of Natural History, 131, 115-210. MacLeod, n. and rose, k. d. 1993. Inferring locomotor behaviour in Paleogene mammals via eigenshape analysis. American Journal of Science, 293A, 300-355. Matthew, w. d. 1901. Additional observations on the Creodonta. Bulletin of the American Museum of Natural History , 14, 1-38. — 1909. The Carnivora and Insectivora of the Bridger Basin, middle Eocene. Memoirs of the American Museum of Natural History, 9, 289-567. 1915. A revision of the lower Eocene Wasatch and Wind River faunas. I. Order Ferae (Carnivora). Suborder Creodonta. Bulletin of the American Museum of Natural History, 34, 4-103. nagurka, m. l. and hayes, w. c. 1980. An integrative graphics package for calculating cross-sectional properties of complex shapes. Journal of Biomechanics, 13, 59-64. o'leary, m. a. and rose, k. d. 1995. Postcranial skeleton of the early Eocene mesonychid Pacliyaena (Mammalia: Mesonychia). Journal of Vertebrate Paleontology, 15, 401-430. polly, p. D. 1996. The skeleton of Gazinocyon vulpeculus n. gen. and comb. (Hyaenodontidae, Creodonta) and the cladistic relationships of Hyaenodontidae (Eutheria, Mammalia). Journal of Vertebrate Paleontology, 16, 303-319. rose, k. d. 1987. Climbing adaptations in the early Eocene mammal Chriacus and the origin of Artiodactyla. Science. 236, 314-316. — 1990. Postcranial skeleton remains and adaptations in early Eocene mammals from the Willwood Formation, Bighorn Basin, Wyoming. 107-133. In bown, t. m. and rose, k. d. (eds). Dawn of the Age of Mammals in the northern part of the Rocky Mountain Interior of North America. Special Paper of the Geological Society of America, 243, I 241. ruff, c. b. and hayes, w. c. 1983. Cross-sectional geometry of Pecos Pueblo femora and tibiae -a biomechanical investigation. I. Method and general patterns of variation. American Journal of Physical Anthropology, 60, 359-381. — walker, a. and teaford, m. f. 1989. Body mass, sexual dimorphism and femoral proportions of Proconsul from Rusinga and Mfangano Islands, Kenya. Journal of Human Evolution, 18, 515-536. runestad, J. a. 1994. Humeral and femoral diaphyseal cross-sectional geometry and articular dimensions in Prosimii and Platyrrhini (Primates) with application for reconstruction of body mass and locomotor behaviour in Adapidae (Primates: Eocene). Unpublished Ph.D. Dissertation, Department of Cell Biology and Anatomy, Johns Hopkins School of Medicine, Baltimore, 443 pp. — ruff, c. b., nieh, j. c., thorington, j. c., jr and teaford, m. f. 1993. Radiographic estimation of long bone cross-sectional geometric properties. American Journal of Physical Anthropology, 90, 207-213. savage, d. e. and russell, d. e. 1983. Mammalian paleofaunas of the world. Addison-Wesley, London, 432 pp. scott, w. b. 1888. On some new and little known creodonts. Proceedings of the Academy of Natural Sciences , Philadelphia, 9, 155-185, 3 pis. smith, r. j. 1984. Allometric scaling in comparative biology: problems of concept and method. American Journal of Physiology, R256, 153-160. springhorn, r. 1980. Paroodectes feisti, der este Miacidae (Carnivora, Mammalia) aus dem Mittel-Eozan von Messel. Paldontologische Zeitschrift, 54, 171 198. — 1982. Neue raubtiere (Mammalia: Creodonta et Carnivora) aus dem Lutetium der grube Messel (Deutschland). Palaeontograpliica, Abteilung A, 179, 105-141. 304 PALAEONTOLOGY, VOLUME 40 — 1985. Zwei neue Skelette von Miacisl kessleri (Mammalia, Carnivora) aus den lutetischen Olschiefern der ‘Grube Messel'. Senckenbergiana Lethaea , 66, 121-141. szalay, F. s. 1977. Phylogenetic relationships and a classification of the eutherian Mammalia. 315-373. In hecht, M. K., goody, p. c. and hecht, B. M. (eds). Major patterns in vertebrate evolution. NATO Advanced Study Institute Series, Series A (14). Plenum Publishing, New York, 450 pp. — and decker, R. L. 1974. Origins, evolution, and function of the tarsus in Late Cretaceous Eutheria and Paleocene primates. 223-259. In jenkins, f. a., Jr (ed. ). Primate locomotion. Academic Press, New York, 390 pp. taylor, m. e. 1974. The functional anatomy of the forelimb of some African Viverridae (Carnivora). Journal of Morphology, 143, 307-336. 1976. The functional anatomy of the hindlimb of some African Viverridae (Carnivora). Journal of Morphology, 148, 227-254. Timoshenko, s. p. and Gere, j. m. 1972. Mechanics of materials. Van Nostrand and Reinhold Co., New York, 541 pp. van valkenburgh, b. 1987. Skeletal indicators of locomotor behaviour in living and extinct carnivores. Journal of Vertebrate Paleontology, 7, 162-182. 1990. Skeletal and dental predictors of body mass in carnivores. 181-205. In damuth, j. and MacFADDEN, b. (eds). Body size in mammalian paleobiology : estimation and biological implications. Cambridge University Press, Cambridge, 397 pp. wang xiaoming 1993. Transformation from plantigrady to digitigrady: functional morphology of locomotion in Hesperocyon (Canidae: Carnivora). American Museum Novitates, 3069, 1-23. wortman, J. L. and Matthew, w. D. 1 899. The ancestry of certain members of the Canidae, the Viverridae, and Procyonidae. Bulletin of the American Museum of Natural History, 12, 109-138. yalden, d. w. 1970. The functional morphology of the carpal bones in carnivores. Acta Anatomica, 77, 481-500. RONALD E. HEINRICH Department of Biological Sciences Ohio University Athens, OH, USA 45701-2579 KENNETH D. ROSE Department of Cell Biology and Anatomy Typescript received 3 January 1996 The Johns Hopkins School of Medicine Revised typescript received 22 August 1996 Baltimore, MD, USA 21205 APPENDIX Abbreviations used for morphological characters discussed in text. ac acromion Met IV fourth metatarsal facet act acetabulum Met V fifth metatarsal facet ah facet for astragalar head nas non-articular shelf ap anconeal process nav navicular facet asf astragalar foramen ntb tubercle of navicular bb bone bridge of olecranon fossa be brachioradialis crest op olecranon process br m. brachialis insertion site pcf posterior calcaneal facet ce capitular eminence pf pit for fibula cen centrale facet Pg groove for peroneal tendons cf coronoid fossa pop nr. popliteus insertion site con phalangeal condyles pt patellar trochlea cp coracoid process ptb peroneal tubercle (process) cty cotylar fossa rf radial fossa cub cuboid facet rn radia notch dl dorsal lip s stop facet for magnum HEINRICH AND ROSE: EOCENE MIACOID CARNIVORANS 305 dp deltopectoral crest sc supinator crest ec articular surface for ectal facet sea expanded articular surface for scaphoid ecf ectal facet sf sustentacular facet ect ectocuneiform facet sgt supraglenoid tubercle ef entepicondylar foramen sm semilunar notch ent entocuneiform facet sn scapular notch exr extensor tubercle of radius ss scapular spine exu extensor tubercle of ungual St scaphoid tubercle ff fibular facet stf supratrochlear foramen till groove for flexor hallucis longus tend. sty styloid process fix flexor tubercle SUS sustentaculum talus fov articular fovea ta m. tibialis anterior fossa gf glenoid fossa tbf tibial facet gtb greater tuberosity tbm tibial malleolus gtr greater trochanter tc tibial crest ib iliac blade tg m. triceps brachii tendinal groove ill iliac tubercle tm m. teres major insertion site itb tubercle for interosseous membrane tint tibial malleolar tubercle it ischial tuberosity tpt m. tibialis posterior tubercle is ischial spine tro astragalar trochlea lpt long plantar tubercle tt tibial tuberosity ltb lesser tuberosity ttr third trochanter hr lesser trochanter tzd trapezoid facet lun lunar facet tzm trapezium facet mag magnum facet unc unciform facet me medial condyle ucl ulnar collateral ligament insertion site me medial epicondyle ul ulnar facet mes mesocuneiform facet ANATOMY AND RELATIONSHIPS OF THE PAREIASAUR PAREIASUCHUS NAS1CORNIS FROM THE UPPER PERMIAN OF ZAMBIA by m. s. y. lee, c. e. gow and j. w. kitching Abstract. A well-preserved and newly prepared skull has enabled a critical re-evaluation of the genus Pareiasuchus (Reptilia; Pareiasauridae) and the species Pareiasuchus nasicornis. The skull is heavily ossified, deeply sculptured, akinetic and anapsid. The skull roof sutures are described for the first time. Most elements typical of basal amniotes are present; however, the postparietal is a single median ossification, the tabular is absent, and a ‘supernumerary element’ (possibly a modified cervical osteoderm) is present. Among the better- known pareiasaur taxa, Pareiasuchus appears to be most closely related to Pareiasaurus , Scutosaurus and Elginia - other forms with highly elaborate cranial ornament. Autapomorphies of Pareiasuchus include : forked distal end of humerus; and proximal end of femoral shaft greatly bent preaxially. Autapomorphies of P. nasicornis include: posteriorly projecting cheek flanges; medially inflected premaxillary and maxillary teeth; and a discrete ossification over each external naris. The specimen demonstrates that many features previously suggested to be unique (amongst basal amniotes) to procolophonoids and turtles also characterize pareiasaurs: for instance, wide antorbital buttress formed by the prefrontal and palatine; short cultriform process; and medially enclosed adductor fossa. P. nasicornis was a large terrestrial herbivore, as evidenced by the crenulated. labiolingually flattened tooth crowns, depressed jaw articulation, small gape, heavy jaws, reduced transverse flange of the pterygoid, and heavy, akinetic skull. The distinctive cheek flange and angular boss were probably defensive structures. Pareiasuchus nasicornis Haughton and Boonstra, 1929 a is one of the most common and distinctive pareiasaurs. Although several complete skulls are known, its cranial anatomy has yet to be adequately described and interpreted. Indeed, the only previous description of the cranium of this taxon is a short paragraph, three photographs of the holotype, and some rough sketches where all the sutures are hypothetical (Boonstra 1934a). The postcranial anatomy, however, has been studied more adequately, and appears to be indistinguishable from that of Pareiasuchus peringueyi (Boonstra 1929a, 19296, 1932, 19346, 1934c; Haughton and Boonstra 1930). Recently, a well-preserved skull of a sub-adult Pareiasuchus nasicornis has been prepared at the Bernard Price Institute of Palaeontology, Johannesburg. This specimen is unusual in that it shows clearly all the external cranial sutures, which are not yet known in Pareiasuchus nasicornis and very poorly known in most other pareiasaurs. A description of this skull should therefore not only fill a gap in our knowledge of the anatomy of Pareiasuchus nasicornis , but also shed light on the systematic position of Pareiasuchus nasicornis within pareiasaurs, and of pareiasaurs within amniotes in general. Recent debate on the last topic (Gauthier et ah 19886; Lee 1993, 1995; Laurin and Reisz 1995) has been hampered by the absence of a good description of even a single pareiasaur. Finally, functional aspects of the highly distinctive pareiasaur skull are discussed. Institutional Abbreviations. BPI, Bernard Price Institute for Palaeontological Research, Johannesburg, South Africa; GSP, Geological Survey, Pretoria, South Africa; PIN. Palaeontological Museum of the Russian Academy of Sciences. Moscow, Russia; SAM, South African Museum, Cape Town, South Africa; WEL, Welwood Museum, Graaff-Reinet, South Africa. | Palaeontology, Vol. 40, Part 2, 1997, pp. 307-335, 1 pl.| © The Palaeontological Association 308 PALAEONTOLOGY, VOLUME 40 SYSTEMATIC PALAEONTOLOGY Remarks. Following recent arguments (e.g. de Queiroz and Gauthier 1992), higher taxa have not been assigned formal Linnean ranks. Rigorous definitions and diagnoses of Amniota, Reptilia, and Pareiasauria are provided in, respectively, Gauthier et al. (1988a, 19886), Laurin and Reisz (1995) and Lee (1995). amniota Haeckel, 1866 reptilia Laurenti, 1768 pareiasauria Seeley, 1888 Genus pareiasuchus Broom and Haughton, 1913 Type species. Pareiasuchus peringueyi. Diagnosis. A moderate-sized pareiasaur, reaching 2 m in snout-vent length, possessing a highly rugose skull, and deep cheek flanges with large pointed bosses on the posterior and ventral edges. Osteoderms are present over the entire dorsum and limbs: most of these are isolated, but some are suturally united. By analogy with Elginia (see Maxwell 1991), the compound osteoderms presumably covered the shoulder and pelvic regions. Pareiasuchus is characterized by numerous unique postcranial specializations (autapomorphies) not found in other pareiasaurs: the ent- and ectepicondyles are both narrow and project distally, the distal end of the humerus therefore appearing ‘forked1; the iliac shaft is greatly inclined anterodorsally, more so than in any other pareiasaur; the second and third sacral ribs are greatly flattened, appearing almost sheet-like; and the proximal end of the femoral shaft is greatly bent preaxially, the femur therefore appearing boomerang-shaped in dorsal or ventral aspect. There are two species, P. peringueyi and P. nasicornis. Remarks. Haughton and Boonstra’s (1929, p. 86) original diagnosis of Pareiasuchus was as follows: ‘snout rounded. Cheek deep and rugose. Ilium as in Pareicisaurus. Ischium long1. This diagnosis is not very adequate, as none of these traits is unique to Pareiasuchus , nor is the combination of all these traits unique. Indeed, all four traits occur in most pareiasaurs. However, the genus Pareiasuchus is monophyletic and can be diagnosed via the other features mentioned above. Pareiasuchus nasicornis Haughton and Boonstra, 1929a Plate 1 ; Text-figures 1-8 1929a Pareiasuchus nasicornis Haughton and Boonstra, p. 86. 19296 Pareiasuchus nasicornis Haughton and Boonstra; p. 263, pi. 36. 1929a Pareiasuchus nasicornis H. and B.; Boonstra, p. 99. 1930 Pareiasuchus nasicornis H. and B.; Haughton and Boonstra, pp. 312, 337, fig. 52. 1932 Pareiasuchus nasicornis H. and B.; Boonstra, pp. 442, 447, 479-481. 1933 Pareiasuchus nasicornis H. and B.; Hartmann-Weinberg, p. 45. 1934a Pareiasuchus nasicornis H. and B.; Boonstra, pp. 8, 23, 33, pis 1-3. 19346 Pareiasuchus nasicornis H. and B.; Boonstra, pp. 44, 48, pi. 27. 1934c Pareiasuchus nasicornis H. and B.; Boonstra, p. 61. 1937 Pareiasuchus nasicornis H. and B.; Hartmann-Weinberg, p. 684. 1940 Pareiasuchus nasicornis H. and B.; Broom, p. 158. 1944 Pareiasuchus nasicornis H. and B.; von Huene, p. 399. 1969 Pareiasuchus nasicornis H. and B.; Kuhn, p. 66, fig. 36; p. 78, fig. 39-7. 1977 Pareiasuchus nasicornis H. and B.; Kitching, pp. 55, 68-69. 1984 Pareiasuchus nasicornis H. and B.; Araujo, pp. 235, 249. LEE ET AL.\ PERMIAN PAREI ASAUR 309 1987 Pareiasuchus nasicornis H. and B.; Ivachnenko, p. 78. 1989 Pareiasuchus nasicornis H. and B. ; Araujo, p. 307. Holotype. SAM 3016: complete skull and lower jaw; parts of both forelimbs, pelvis, and both hindlimbs; numerous osteoderms. Type locality and horizon. Graaff-Reinet Commonage, South Africa; Cistecephalus Zone, Upper Permian (Kitching 1977). Referred specimens. BP1 1/254: skull fragment and partial lower jaw. BPI 1/1500: antorbital region of skull, with mandible missing part of left ramus. BPI 1/3653: complete skull and lower jaw; unprepared blocks containing anterior cervicals, osteoderms and portions of (?) shoulder girdle. GSP 475; complete skull and lower jaw; posterior 3 sacrals and anterior caudals; portions of shoulder and pelvic girdles. GSP CBT4: skull missing most of roof; complete lower jaw. GSP R320: pelvis; portions of both forelimbs and both hindlimbs; 1 7 caudals ; ribs and osteoderms. GSP TN257 : complete skull and lower jaw. SAM K6607 : complete skull with lower jaw; unprepared postcranial blocks containing vertebrae, parts of pelvis and hindlimb, and osteoderms. All except BPI 1/3653 are from the Cistecephalus Zone of South Africa. BPI 1/3653 (described here) is from the Cistecephalus Zone of Zambia. Diagnosis. Based on previously studied specimens, P. nasicornis differs from the very similar P. peringueyi in lacking the two autapomorphies for that taxon: the blunt angular boss, and long posteriorly directed spine on the palatal ramus of the quadrate (Haughton and Boonstra 19296; Lee in press). It also possesses at least two autapomorphies of its own: the large descending cheek flanges project backwards, making the skull appear ‘delta-shaped’ in dorsal aspect; and the marginal teeth in the upper jaw point inwards towards the palate. In all other pareiasaurs (including P. peringueyi ), the large cheek flanges, when present, project laterally. In most other pareiasaurs (including Pareiasuchus peringueyi) the marginal teeth in the upper jaw point directly downwards. However, inflected marginal teeth occur in one other pareiasaur, Bradysaurus baini. The present study (see below) has revealed two other autapomorphies of P. nasicornis: the anteriorly directed boss over the external nostril is a discrete ossification, not an extension of the nasal bone; and the cultriform process is shorter and wider than in all other pareiasaurs. Remarks. P. nasicornis might possess two other autapomorphies. In the femur of adults (SAM 3016 and GSP R320), the internal trochanter extends distally all the way to the preaxial tibial facet. However, this condition does not occur in a juvenile (SAM K6607), where the internal trochanter fades out along the shaft of the femur, the condition found in all other pareiasaurs including P. peringueyi. As befits its name, P. nasicornis possesses distinct, anteriorly directed bosses over the external nares, close to the skull midline (Haughton and Boonstra 1929a). These are poorly developed in P. peringueyi. However, among other pareiasaurs, the bosses are also well-developed in a few individuals of Scutosaurus (Ivachnenko 1987). They are also well-developed in Elginia , but are located further laterally, away from the skull midline (Newton 1893; Walker 1973; Maxwell 1991). The distinct, sagitally located bosses over the nares in P. nasicornis are probably autapomorphic for this species, occurring convergently in the other two taxa, but this interpretation must remain tentative. Haughton and Boonstra (1929a), followed by Boonstra (1934a), suggested that Pareiasuchus nasicornis further differed from P. peringueyi in possessing smaller bosses on the skull roof and cheek, and a shorter palate. However, there appear to be no clear-cut differences in these areas. The bosses are weakly developed in juveniles of both Pareiasuchus nasicornis (e.g. GSP TN257) and P. peringueyi (e.g. WEL RC784), and are prominent in adults. The palate is of similar shape in both species. Haughton and Boonstra (1929a) suggested that the angular boss in P. nasicornis was inflected medially, unlike that in any other pareiasaur. However, this appears to be a taphonomic artefact : in their specimen (SAM 3016), the boss appears to have been bent inwards by dorso-ventral crushing. 310 PALAEONTOLOGY, VOLUME 40 DESCRIPTION General Material (PI. 1; Text-figs 1-2). BPI 1/3653, complete skull and lower jaw; plus a small unprepared block containing disarticulated anterior cervicals, osteoderms and portions of shoulder girdle. As discussed above, the postcranial anatomy of P. nasicornis has already been described, and this work will therefore concentrate on the cranium. Locality. Northern (Chikonta) Group, Locality 21 of Kitching (1963), Upper Luangwa Valley, Zambia (formerly Northern Rhodesia). Horizon. Lowermost Cistecephalus Zone. The genus Cistecephalus occurs at locality 21 as isolated specimens together with abundant small and some medium-sized dicynodonts (Drysdall and Kitching 1963; Kitching 1963). This faunal assemblage is approximately equivalent to the original Endothiodon Zone of Broom (1909), renamed the Tropidostoma-Endothiodon Assemblage Zone by Keyser and Smith (1977). This horizon has been incorporated into the lowermost Cistecephalus Zone (Kitching 1970, 1977). Identification. BPI 1/3653 can be referred to Pareiasuchus nasicornis on the basis of possession of all three cranial autapomorphies already known to be characteristic of that species: the backwardly projecting cheek flanges, inflected marginal teeth on the upper jaw, and prominent nasal bosses. It is identical to the type and to all other referred specimens of P. nasicornis in almost all other features, although each of these traits is also found in at least some other pareiasaurs. There are slight differences in the cranial ornamentation among all the skulls referred to P. nasicornis but, as suggested in the discussion below, these are probably ontogenetic differences. Soft anatomy. Information on soft anatomical features associated with various osteological landmarks (foramina, grooves, etc.) is derived from Gaffney (1979, 1990) and Heaton (1979). These are the two most comprehensive published discussions of basal amniote soft anatomy. Heaton used extant diapsids as analogues for Eocaptorhinus , whereas Gaffney (1979, 1990) used extant turtles as analogues for extinct turtles such as Proganochelys. However, for all the osteological landmarks discussed in this work, the two workers reached identical conclusions regarding the associated soft anatomy, suggesting that their conclusions are general for basal amniotes, including pareiasaurs such as Pareiasuchus. In particular, because turtles and captorhinids form successive extant outgroups to the taxon under study, soft anatomical features common to both can be reasonably inferred to characterize P. nasicornis as well, unless the osteology of P. nasicornis suggests otherwise (Bryant and Russell 1992; Witmer 1995). Dermal roofing elements (PI. 1 ; Text-figs 1-2, 4-5) General. The elements of the skull roof are well-preserved. The posterior part of the skull table is slightly damaged. Most of the posterior edge is weathered, and a portion on the right side is missing. Consequently, the posterior limit of the postparietal and supernumerary elements cannot be ascertained, and the dorsal margin of the posttemporal fenestra, formed by the supernumerary and supratemporal elements, is unclear. A narrow wedge-shaped strip is missing from the left cheek behind the orbit, but this region is preserved on the right side. The shape of the pineal foramen indicates that the skull table has been slightly compressed anteroposteriorly. Also, comparisons with other skulls of P. nasicornis (see Referred specimens ) indicate that the right cheek is undistorted, while the left cheek has been pushed slightly forwards. This is consistent with the observation that the right cheek is undamaged whereas the left cheek has a portion missing. EXPLANATION OF PLATE 1 Pigs 1-2. Pareiasuchus nasicornis Haughton and Boonstra, 1929a; BPI 1/3653; skull. 1, in dorsal view; 2, in ventral view; xO-4. PLATE 1 LEE et a l Pareiasuchus 312 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Pareiasuchus nasicornis Haughton and Boonstra, 1929a; BPI 1/3653; skull in a, left lateral view; b, right lateral view; xO-42. The skull is approximately triangular in dorsal (or ventral) aspect, with a broad, rounded snout and large, posterolaterally directed cheek flanges. In lateral view, the snout is the same height along its entire length; it does not taper anteriorly. The external naris is a large, circular opening that faces anteriorly. The orbit faces laterally and is located very slightly in front of the middle of the skull. It is elongated along the anterodorsal- LEE ET AL.\ PERMIAN PAREI ASAUR 313 posteroventral axis. The pineal foramen is located slightly behind the posterior limit of the orbit. Unusually for pareiasaurs, almost all external cranial sutures are clearly visible. Most roofing elements characteristic of basal amniotes are present. However, the postparietal is a single median ossification, and the tabular is absent (the element sometimes identified as the ‘tabular’ is the supratemporal : see Discussion). A supernumerary element is present between the postparietal and the supratemporal. B text-fig. 2. Pareiasuchus nasicornis Haughton and Boonstra, 1929a; BPI 1/3653. A, skull in posterior view; a block of unprepared postcranial fragments, outlined in a thick black border, partly obscures the right side of the specimen in this photograph; xO-53. b, view looking forwards and medially into the left orbit, showing details of the antorbital buttress; x 0-87. 314 PALAEONTOLOGY, VOLUME 40 The external surface of each bone on the skull roof, except the premaxilla and maxilla, is heavily sculpted. Usually, one or more bosses are located near the centre of the element, and irregular, rugose ridges radiate over the rest of the external surface. Small rounded pits are present in the grooves between the ridges. Nasal. The nasal is a quadrilateral element which tapers anteriorly and is slightly longer than wide. It sutures with its partner along the skull midline. A short groove on is external surface, near its anterior tip, receives the dorsal process of the premaxilla. The nasal also contacts the lacrimal, prefrontal, and frontal, and forms the dorsal margin of the external naris. Two bosses are present on the anterior part of each nasal. On the anterior tip of each nasal, directly above the suture with the premaxilla, there is another boss composed of a discrete ossification. This boss, and the suture, are clearly visible on the left nasal. On the right nasal, the boss is missing, and the sutural surface of the nasal is exposed. Frontal. The frontal is an approximately rectangular element forming most of the interorbital region. Medially, it has a long suture with its partner along the skull midline. It also contacts the nasal, prefrontal, postfrontal and parietal. It does not enter the orbital margin. A prominent boss is present, located slightly towards the rear of the element. Parietal. The parietal is a large, squarish element forming most of the posterior skull roof. It sutures with the frontal, postfrontal, postorbital, supratemporal and postparietal. Medially, it forms a long suture with its partner along the skull midline. The pineal foramen is located along this suture, about one-third of the way back from the anterior end. In all other skulls of P. nasicomis, the foramen is round: in this specimen anteroposterior compression has distorted it into a transverse slit. A large, rounded boss is present in the centre of the parietal. Postparietal. The postparietal is the only unpaired element on the skull roof. It is a median, U-shaped bone that forms the centre of the posterior part of the skull table. Anteriorly, it extends into an embayment between the parietals. Laterally, it sutures with the supernumerary element. The external surface of the postparietal is covered with the usual rugae and pits; however, there is no central boss. A sagittal process on the internal (ventral) surface of the postparietal meets the dorsal process of the supraoccipital, forming a robust pillar connecting the braincase to the skull table. The posterior margin is damaged and has been reconstructed on the basis of other specimens of P. nasicomis (e.g. SAM 3016). Supernumerary element. The supernumerary element is a small triangular bone wedged between the postparietal, parietal, and supratemporal. The external surface lacks a boss, but is weakly sculpted with the usual rugae and pits. The posterior margin is damaged and has been reconstructed through comparison with other specimens of P. nasicomis (e.g. SAM 3016). The homology of this bone is addressed in the Discussion below. Premaxilla. The premaxilla is a small element forming the tip of the snout and the medial and ventral margins of the external naris. Medially, it sutures with its partner along the skull midline. The premaxilla also sutures with the maxilla and nasal. The palatal portion of the premaxilla is obscured by the lower jaws; however, in all other pareiasaurs, the premaxilla sutures with the vomer, and forms the anterior margin of the internal naris and the lateral and anterior margin of the single median foramen praepalatinum. The premaxilla lacks dermal ornament and bears two large alveoli. Maxilla. The maxilla is a vertical plate of bone, highest anteriorly and tapering posteriorly. Anteriorly, it forms most of the posterior margin of the external naris. It does not enter the orbital margin. On the skull roof, the maxilla sutures with the premaxilla, lacrimal, and jugal. The posterior limit of the maxilla just contacts the anterior tip of the quadra tojugal, excluding the jugal from the ventral cheek margin. The entire external surface of the maxilla is unornamented. Anteriorly, there is a foramen maxillare, representing the exit for a ramus of the arteria alveolis superior. Another, smaller foramen further back along the external surface of the maxilla may represent the exit of another branch of this artery. A groove for the blood vessel extends anteriorly from each of these openings. The ventral surface of the palatal portion of the maxilla is obscured by the articulated lower jaws in all specimens of P. nasicomis. However, part of the dorsal surface is visible through the orbit of BPI 1/3653 (PL 1. fig. 1 ; Text-fig. 4a). A narrow horizontal palatal ledge is present on the medial surface of the maxilla. This LEE ET AL.\ PERMIAN PAREIASAUR 315 text-fig. 3. Pareiasuchus nasicornis Haughton and Boonstra, 1929m a, anterior view of snout, showing premaxillary and anterior maxillary teeth in labial view. The black vertical line is the skull midline suture, between the premaxillae and the dentaries; x 103. b, lingual view of partially erupted tooth from the middle of the mandibular tooth row. Because of difficulty of access, preparation and photography of this tooth has been poor; x 0 94. ledge sutures with the ectopterygoid and palatine. The area of the ledge in front of the orbit is obscured by the antorbital buttress. Presumably, as in all other pareiasaurs, the ledge continues anteriorly in front of the palatine and forms part of the lateral border of the internal nans. The maxilla bears 12 alveoli, which are largest anteriorly and gradually decrease in size posteriorly. The alveolar ridge is narrow, barely wider than the tooth bases. Lacrimal. The lacrimal is a long element forming much of the facial region, much of the anterior margin of the orbit and a very small portion of the margin of the external naris. It is largely excluded from the latter opening by the anterior expansion of the maxilla, and sutures with the nasal, prefrontal, maxilla and jugal. Along the orbital margin, the internal surface of the lacrimal sutures with the descending (antorbital) flange of the prefrontal. The foramen orbitonasale penetrates the base of the antorbital buttress, at the quadruple junction between the lacrimal, prefrontal, palatine and jugal. This opening between the fossa orbitalis and the fossa nasalis transmitted the palatine branch of the facial nerve (VII), the lateral ramus of the facial nerve, and the prefrontal vein. A single large rounded boss is present, located slightly towards the anterior of the element. Prefrontal. The prefrontal consists of a curved plate which forms part of the skull roof, and an antorbital buttress. The skull roof portion is horizontal and sutures with the lacrimal, nasal, frontal, and postfrontal. Two prominent bosses are located above the orbital margin. The antorbital buttress is a curved, vertical flange (Text-fig. 5b). It is oriented anterolaterally and is narrowest dorsally, gradually widening ventrally. The medial edge is smoothly concave; the lateral edge sutures with the medial surface of the lacrimal. Ventrally, the descending flange of the prefrontal meets a dorsal flange of the palatine and forms part of the medial border of the foramen orbitonasale (see Lacrimal). Postfrontal. The postfrontal is a sizeable, semicircular element which forms most of the dorsal orbital margin. It sutures with the prefrontal, frontal, parietal, and postorbital, and bears a prominent boss nears its anterolateral edge. 316 PALAEONTOLOGY, VOLUME 40 text-fig. 4. For caption see opposite. LEE ET AL.\ PERMIAN PAREIASAUR 317 text-fig. 5. Pareiasuchus nasicornis Haughton and Boonstra, 1929a. Reconstruction of the skull in a, left lateral, and b, occipital view. Scale bars represent 30 mm. Postorbital. The postorbital is a curved plate, consisting of a horizontal dorsal portion, which forms part of the skull table, and a ventrolaterally oriented ventral portion, which forms part of the cheek. It forms much of the posterodorsal portion of the orbital margin, and sutures with the prefrontal, parietal, supratemporal. text-fig. 4. Pareiasuchus nasicornis Haughton and Boonstra, 1929a. Reconstruction of the skull in a, dorsal, and b, ventral view. Scale bar represents 30 mm. 318 PALAEONTOLOGY, VOLUME 40 squamosal, and jugal. A prominent boss is present near the orbital margin, on the junction between the cheek and skull table. Supratemporal. The supratemporal is an irregularly shaped plate which forms most of the posterior skull roof. It sutures with the parietal, supernumerary, supratemporal, squamosal, and postorbital. Most of the external surface is heavily sculpted, but this ornamentation becomes weaker near the posterior edge. There is a prominent central boss. A ventrally directed flange is present on the medial surface of supratemporal, near the posterolateral corner. This flange meets the upturned distal end of the paroccipital process. Squamosal. The squamosal is a flat, rectangular element that forms the posterodorsal region of the cheek. It sutures with the supratemporal, postorbital, jugal, and quadratojugal. Three prominent central bosses are present on the external surface. On the medial surface, a vertical flange sutures with the dorsal ramus of the quadrate. The posterior edge of the squamosal forms part of the posterior margin of the cheek. Along this edge, there is a peculiar raised area of uncertain function. This area is teardrop-shaped, with the pointed end directed dorsomedially. The surface of this raised area is shallowly concave. Immediately below this structure, a narrow groove extends dorsolaterally across the posterior edge of the cheek. Jugal. The jugal is a flat, crescent-shaped element that forms the cheek region below and behind the orbit. It forms the entire ventral margin of the orbit, and sutures with the postorbital, squamosal, quadratojugal, maxilla, and lacrimal. A central clump of bosses is present, below the posteroventral margin of the orbit. The ventral surface of the palatal portion of the jugal is obscured by the articulated lower jaw in all specimens of P. nasicornis. However, the dorsal surface is visible through the orbit of BPI 1/3653 (PI. 1, fig. 1 ; Text-fig. 2a): there appears to be a small horizontal palatal flange which extends medially to overlap the ectopterygoid. This has been termed the ‘medial process’ (Gaffney and Meylan 1988; Gaffney 1990), the ‘alary process’ (Gauthier et al. 1988 b), or the ‘internal process' (Heaton 1979). Quadratojugal. The quadratojugal is the largest element on the skull roof. It is a flat, quadrilateral bone that forms most of the ventral portion of the cheek. The anterior tip of the quadratojugal just manages to contact the maxilla. The quadratojugal also sutures with the jugal and squamosal. The quadratojugal is greatly expanded ventrolaterally, far beyond the level of the tooth row. This expansion forms a large cheek plate that covers the posterior half of the lower jaw, and is the structure upon which the name ‘pareiasaur’ (cheek-lizard) is based (Owen 1876). The ventral corner of this cheek plate bears a very prominent conical boss, the posterior margin is thickened and adorned with several smaller bosses. Another three small bosses are present immediately in front of the large corner boss; however, further anteriorly, the edge of the cheek plate is a straight, thin, unornamented edge. This edge meets the alveolar ridge of the maxilla at an angle of about 135°. A flange extends medially from the internal surface of the quadratojugal, to meet the quadrate (Text-fig. 5b). Dorsally, this flange is continued by a similar structure on the squamosal. The quadrate foramen is located on the triple junction of the quadratojugal, squamosal, and quadrate (see Quadrate). Palate , palatoquadrate and braincase (PI. 1, fig. 2; Text-figs 2, 4b, 5b, 6) General. The braincase and palatal elements are well-preserved, with only slight distortion. Parts of the braincase are covered by the skull roof and much of the lateral and anterior regions of the palate are obscured by the articulated lower jaws. The basioccipital has been pushed slightly dorsally towards the skull roof, and the distal ends of both paroccipital processes have partially separated from the skull roof. The missing portion of the skull table has exposed the dorsal surface of the paroccipital process and lateral surface of the supraoccipital. The stapes is not preserved; the copula (basihyal), and the left second ceratobranchial, are present but have been displaced towards the right ramus of the lower jaw. The palate forms a horizontal roof over the oral cavity, and is raised well above the level of the alveolar ridge. All the elements characteristic of basal amniotes are present, including the ectopterygoid. The internal naris is large, and the posterior portion is located well away from the alveolar ridge. The foramen palatinum posterius is large, and is also located well away from the alveolar ridge. A small interpterygoid vacuity is LEE ET AL. PERMIAN PAREIASAUR 319 text-fig. 6. Pareiasuchus nasicornis Haughton and Boonstra, 1929m a, braincase of BPI 1/3653, in left lateral view, b, reconstruction of the braincase in left lateral view, using information derived from both sides of the specimen. The paroccipital process and the left cheek are omitted, c, reconstruction of basisphenoid and basioccipital in ventral view. Scale bar represents 20 mm. present, and the basipterygoid ‘articulation’ is sutural and completely immobile. The transverse flange of the pterygoid is weakly developed and the subtemporal fenestra is consequently extensive. A single row of small conical denticles extends along the posterior edge of the transverse flange of the pterygoid. A double row extends anterolaterally from the interpterygoid vacuity towards the foramen palatinum posterius. Another double row extends anteriorly from the interpterygoid vacuity to the posterior margin of the internal naris. This double row continues anteriorly, medial to the internal naris, as a single row of larger denticles, and posterolaterally along the margin of the interpterygoid vacuity as a single row of small denticles. Vomer. The vomer is a long element that forms most of the palate medial to the internal naris. Medially, it sutures with its partner along the skull midline. The anterior portion of the vomer is obscured by the articulated lower jaws in all individuals of P. nasicornis. Presumably, as in all other pareiasaurs, it forms the posterior border of the single median foramen praepalatinum, and contacts the palatal process of the premaxilla. The lateral edge of the vomer forms the medial margin of the internal naris. This margin is concave both anteriorly and posteriorly. The posterior limit of the vomer is unclear in all specimens of P. nasicornis. Presumably, as in Scutosaurus karpinskii , the only pareiasaur in which this suture is known (Lee 1994b), the vomer extended slightly behind the posterior limit of the internal naris, and sutured with the palatine and pterygoid. A single, high, longitudinal ridge, bearing a row of approximately nine large conical denticles, extends along the medial margin of the vomer. This ridge is located on the anterior half of the vomer, very close to the 320 PALAEONTOLOGY, VOLUME 40 midline. This ridge continues posteriorly along the vomer and on to the pterygoid, as a weaker ridge with smaller denticles. Near the presumed pterygoid-vomer suture, another similar denticulated ridge begins. This ridge is positioned lateral and parallel to the previous ridge, and also continues posteriorly along the pterygoid. Palatine. The palatine is an irregularly shaped element that forms much of the lateral region of the palate, between the internal naris and the foramen palatinum posterius. A long, tapering process extends anteriorly, between the internal naris and the maxilla. The ventral surface of the palatine-maxilla suture is obscured by the lower jaw in all specimens of P. nasicornis. The palatine also sutures with the ectopterygoid, pterygoid, lacrimal, and presumably the vomer (see Vomer). A large foramen palatinum posterius occurs in the middle of the palatine-ectopterygoid suture. This opening is oval, being slightly elongated along the suture. In turtles and squamates, this foramen is mostly filled with connective tissue, but also forms the ventral exit for branches of the nasal and maxillary arteries, and of the maxillary nerve (Gaffney 1979, 1990). On the ventral surface, medial to the foramen palatinum posterius, two ridges extend posteriorly and slightly medially. Each bears 3 or 4 small conical denticles. Anteriorly, these ridges converge slightly; posteriorly, they continue on to the pterygoid. On the dorsal surface of the palatine, in front of the orbit, a flange extends perpendicularly upwards to meet the descending flange of the prefrontal and enter the margin of the foramen orbitonasale (see Prefrontal). Ectopterygoid. The ectopterygoid is a small rectangular element that forms the portion of the palate behind the foramen palatinum posterius. It contacts the maxilla, jugal, pterygoid, and palatine. The foramen palatinum posterius (see Palatine) occurs in the centre of the suture with the palatine. The ventral surface of the lateral portion of the ectopterygoid is obscured by the lower jaws in all specimens of P. nasicornis : however, the dorsal surface is visible through the orbit of BPI 1 /3653. The posterolateral edge of the ectopterygoid is concave and forms the anterior margin of the subtemporal fenestra. The exposed area of the ventral surface of the ectopterygoid does not bear any ridges or denticles. Pterygoid. The pterygoid is a large, complex element. Anteriorly, it sutures with the palatine and ectopterygoid. The suture with the vomer is unclear, but appears to be located at the level of the posterior margin of the palatine. Laterally, the pterygoid forms the medial border of the subtemporal fenestra. The transverse flange (processus pterygoideus externus) is weakly developed. It extends horizontally, with only a slight ventrolateral inclination, and therefore does not project ventrally below the tooth row. The posterior border of the transverse flange bears a single row of approximately 13 small conical denticles. The pterygoids are sutured together along the anterior midline. Posteriorly, a small triangular interpterygoid vacuity is present between them. The anterior border of this vacuity is rounded. A ridge, bearing approximately 12 small conical denticles, extends anteriorly from the anterior end of the interpterygoid vacuity. Another ridge, bearing approximately 20 denticles, lies parallel and immediately lateral to it and continues posterior along the lateral margin of the interpterygoid vacuity. Both ridges continue anteriorly onto the vomer. Another pair of ridges extends anterolaterally, from the middle of the pterygoid towards the foramen palatinum posterius. Each ridge bears approximately 15 small conical denticles. Posteriorly, the ridges converge. Anteriorly, the ridges continue onto the palatine. Behind the interpterygoid vacuity, the pterygoid forms an extensive suture with the basipterygoid process of the basisphenoid. The plane of the sutural surface faces dorsally and slightly posteriorly and medially. The basicranial 'articulation' is therefore completely akinetic. The quadrate ramus of the pterygoid extends posteriolaterally from the basicranial articulation, and forms most of the wall separating the temporal fossa from the middle ear. The wall is lowest medially, near the basicranial articulation. Laterally, the quadrate ramus gradually gets higher until, near the cheek, it almost reaches the paroccipital process. Further laterally, the wall is continued by the quadrate. The suture between the quadrate and pterygoid is, however, unclear in all known specimens of P. nasicornis , and has been restored according to the pattern found in P. peringueyi (BPI 1/4105). From the ventral margin of the quadrate flange, a wide horizonal ledge extends posteriorly, partly flooring the cranioquadrate space. This ledge continues medially on to the basipterygoid process, and laterally on to the quadrate. Quadrate. The quadrate is a triangular, vertical plate, oriented perpendicularly to the skull midline. It forms part of the wall between the middle ear cavity and the subtemporal fossa. Medially, the quadrate contacts the quadrate flange of the pterygoid, although this suture is unclear (see Pterygoid). The lateral margin contacts the flange on the medial surface of squamosal and the quadratojugal. The quadrate foramen is located at the LEE ET AL.\ PERMIAN PARE1ASAUR 321 triple junction of the quadrate, squamosal, and quadratojugal. This foramen has not been reported in turtles, but is present in captorhinids and diapsids, where it is traversed by the mandibular nerve and ‘muscular artery' (Heaton 1979). The dorsal tip of the quadrate extends almost to the distal end of the paroccipital process. From the ventrolateral corner of the quadrate, a descending process bears the articular surface for the lower jaw. This surface is obscured by the articulated lower jaws in all individuals of P. nasicornis. Presumably, as in other pareiasaurs, it consists of a lateral and a medial condyle, separated by a groove. The posterior surface of the quadrate forms part of the anterior wall of the middle ear cavity. The ventral margin bears a horizontal ledge, which projects posteriorly and partially floors this cavity. This ledge bears a small tubercle and continues medially onto the pterygoid. The tubercle is not found in captorhinids or turtles, but is present in some other basal amniotes (e.g. procolophonoids). No function has yet been proposed for it, although it might have been part of the articulation with the stapes. Proganochelys and captorhinids both have a pit in exactly this position, the columellar recess (Heaton 1979) or ‘quadrate pocket' (Gaffney 1990), which has been interpreted to receive the distal end of the stapes. Basisphenoid. The area occupied by the parasphenoid and basisphenoid in primitive tetrapods is occupied by a single complex ossification, here termed simply the basisphenoid, following Gaffney (1979, 1990). This element forms the anterior portion of the thick floor of the cavum cranii. Posteriorly it sutures with the basioccipital. The sutural surface is deeply concave transversely, and faces posterodorsally. Thus, the basioccipital extends under (i.e. ventral to) the basisphenoid, and forms much of the external ventral surface of the braincase. Posteriorly, two large hemispherical basal tubera occur on this surface, along the basisphenoid-basioccipital suture. On the lateral surface, the basisphenoid-basioccipital junction appears as a curved, anterodorsally oriented suture which extends to the ventral margin of the fenestra ovalis. The basisphenoid forms the anteroventral margin of the fenestra ovalis. In front of the fenestra ovalis, the basisphenoid forms the base of the lateral wall of the cavum cranii. This wall is continued dorsally by the prootic. The basipterygoid process projects anterolaterally from the anterior of the basisphenoid, and is firmly sutured to the pterygoid. The sutural area is very long (anteroposteriorly) and faces ventrally, and slightly anteriorly and laterally. The basisphenoid extends anteriorly as a wide, extremely short cultriform process. It contains two foramina on the ventral surface, near the basipterygoid process. The posterior foramen exits on the dorsal surface of the basipterygoid process; the course of the anterior foramen is not known. These openings have not been reported in any other basal amniotes, and their former contents are uncertain. Prootic. The prootic forms part of the anterior sidewall of the cavum cranii. Anteriorly, its dorsal margin is embayed for the trigeminal nerve, while posteroventrally it forms part of the border of the fenestra ovalis. The facial (VII) nerve exits from a small foramen in the centre of the prootic. The prootic sutures with the basisphenoid ventrally, the opisthotic posteriorly, and the supraoccipital dorsally. Opisthotic. The opisthotic forms most of the posterior sidewall of the cavum cranii and most of the paroccipital processes. An anteroventral flange forms the posterior margin of the fenestra ovalis. Ventrally, the opisthotic forms the dorsal border of the very large foramen jugulare anterius, traversed by the vagus (X) and accessory (XI) nerves, and the vena cerebralis posterior. The fenestra ovalis and the foramen jugulare anterius are confluent, although almost separated by processes from the opisthotic and basisphenoid. Posterior to the foramen jugulare anterius, the opisthotic sutures with the exoccipital, which sends a long flange laterally along the ventroposterior surface of the paroccipital process. Dorsally, above the paroccipital process, the opisthotic sutures with the supraoccipital. The paroccipital process is situated immediately above the fenestra ovalis and foramen jugulare anterius. It is a thick, curved plate of bone that extends posterolaterally to the cheek. The transverse axis is also slightly inclined, so that the dorsal surface faces slightly posteriorly, while the ventral surface faces slightly anteriorly. The dorsal surface is convex, and the ventral surface flat, along the anteroposterior dimension. Proximally, the paroccipital process is very broad anteroposteriorly. Distally, it gradually becomes narrower. The distal tip curves dorsally and sutures with the supratemporal and squamosal. In this specimen, however, distortion has caused the paroccipital process and the skull roof to separate slightly. Exoccipital. The exoccipital is a small element forming the posterior region of the braincase and part of the occiput. It consists of a vertical plate, which forms the lateral wall of the posterior region of the cavum cranii and part of the occipital condyle, and a horizontal flange, which extends along the paroccipital process. 322 PALAEONTOLOGY, VOLUME 40 Anteriorly, the vertical plate forms the posterior border of the foramen jugulare anterius. Above this foramen, it has a horizontal suture with the opisthotic. Below this foramen it has a horizontal suture with the basioccipital. Posteriorly, it forms the lateral and ventral borders of the foramen magnum. Above the foramen magnum, the exoccipital sutures with the supraoccipital : it does not meet its partner dorsally. The foramen magnum is oval, being widest along the horizontal axis. Below the foramen magnum, the exoccipital sutures with its partner and projects posteriorly beyond the level of the foramen magnum. This portion forms the laterodorsal corner of the occipital condyle. The two exoccipitals are almost completely separated by the basioccipital, but just contact one another along the dorsal margin of the occipital condyle. Lateral to the foramen magnum, the exoccipital sends a long triangular flange along the paroccipital process. This flange covers the proximal portion of the posterior surface of the paroccipital process, tapering distally. It also covers the ventral surface of the paroccipital process near the foramen magnum. Basioccipital. The basioccipital forms the thick posterior floor of the cavum cranii, and most of the occipital condyle. It extends anteriorly over the dorsal surface of the basisphenoid. In lateral view, basisphenoid- basioccipital suture extends posteroventrally. In ventral view, the basioccipital-basisphenoid suture is deeply convex anteriorly: the basioccipital projects anteriorly between the basal tubera. Dorsally, the basioccipital forms the ventral margin of the foramen jugulare anterius (see Opisthotic). The suture with the exoccipital extends posteriorly from the foramen to the surface of the occipital condyle. The basioccipital forms most of the occipital condyle, except for the dorsolateral portions which are formed by the exoccipitals. The condyle is almost circular, but the dorsal edge, under the foramen magnum, is slightly flattened. The articulatory surface is concave, and projects a short distance behind the foramen magnum. Supraoccipital. The supraoccipital is a thick vertical pillar. Dorsally, it sutures with the median ventral flange from the postparietal. Ventrally, the supraoccipital is expanded, forming the roof of the posterior portion of the cavum cranii. Laterally, this expansion sutures with the opisthotic, above the paroccipital process. Posteriorly, it projects between the two exoccipitals, forming the dorsal margin of the foramen magnum. Pleurosphenoid. The region of the pleurosphenoid is inaccessible and this element has not been prepared. It was presumably present, as in all other pareiasaurs. Stapes (Columella auris). The stapes is not preserved. Hyoid apparatus. (Text-fig. 8c). The copula (basihyal) is a long, flat plate, slightly expanded at each end for articulation with the ceratobranchials. The anterior edge is firmly pressed against the lower jaw and thus not accessible to preparation. This suggests that the lingual process is absent, and that the anterior edge was most probably straight, as in all other pareiasaurs. The posterior edge is concave. The second ceratobranchial is a long bent rod, widest proximally and gradually tapering distally. Lower jaw (PI. 1, fig. 2; Text-figs 1, 4b, 7) General. The lower jaw is preserved in articulation, and is complete and undistorted. However, the jaws are tightly closed, making certain areas inaccessible to preparation. All the mandibular elements characteristic of basal amniotes are present. There is only one coronoid, as in all basal amniotes except synapsids. The surfaces of the lower jaw are smooth and unsculpted. Each ramus is short and very deep. The posterior portions of the rami are straight and almost parallel, converging very slightly; anteriorly, they curve medially to meet each other and form a broad, smoothly rounded arch. The lower jaw is slightly smaller than the upper jaw, so that the labial surfaces of the mandibular teeth shear past the lingual surfaces of the maxillary and premaxillary teeth. The articular cotyle is positioned slightly below the level of the tooth row. Dentary. The dentary forms most of the anterior portion of the lower jaw. It is exposed extensively on the lateral surface, overlapping the splemal anteroventrally, the angular posteroventrally and the surangular posterodorsally. Anteriorly, it sutures with its partner and forms the dorsal half of the symphysis. On the medial surface, the dentary is almost completely covered by the splenial : only the anterior tip, and a thin splint above the splenial, are exposed. The alveolar ridge is inaccessible to preparation. In all other pareiasaurs, the lower jaw has slightly fewer teeth than the upper jaw, and the dentary teeth decrease in size posteriorly; based on this, there were probably about 12 alveoli on the dentary, decreasing in size posteriorly. LEE ET A L : PERMIAN PAREIASAUR 323 fna *dm text-fig. 7. Pareiasuchus nasicornis Haughton and Boonstra, 1929n. Reconstruction of the lower jaw in a. lateral view, b, medial view, c, ventral view. Scale bar represents 30 mm. Two foramina are visible on the lateral surface of the dentary. A small unnamed opening is located half-way up the lower jaw, near the symphysis, while a slightly larger opening, the foramen dentofaciale majus, is located slightly more posteriorly and dorsally. The former contents of these foramina are, surprisingly, unknown (see Gaffney 1979). The dentary forms the dorsal wall of a groove on the medial surface of the lower jaw, the sulcus cartilaginis meckelii, occupied by part of Meckel’s cartilage. This groove extends horizontally from the symphysis and leads posteriorly to the foramen intermandibularis medius, which contained Meckel's cartilage and the ramus intermandibularis medius of the mandibular (V3) nerve. The lateral edge of this foramen is formed by the dentary, the medial edge by the splenial. Angular. The angular is mostly covered by the other elements of the lower jaw. Its greatest exposure is ventral, where it forms the posterior two-thirds of the ventral surface of the mandible. On the lateral surface, the angular is overlapped by the dentary anteriorly and by the surangular dorsally. Posteriorly, it covers the ventral surface of the articular. On the medial surface, the angular is overlapped by the splenial anteriorly and the prearticular dorsally. The angular forms the ventral half of the border of the foramen intermandibularis caudalis, a large oval opening situated mid-way along the angular and traversed by the ramus intermandibularis caudalis of the mandibular (V3) nerve. Anterior to this foramen, on the ventral surface of the angular, there is a large rounded swelling, the ‘angular boss’. Surangular. The surangular is a shcet-like element that forms much of the lateral surface of the posterior portion of the lower jaw. It overlaps the angular ventrally, the articular posteriorly, and is overlapped by the dentary anteriorly. Its dorsal edge forms the lateral rim of the adductor fossa. In the centre of the surangular, there is a small round opening, the foramen nervi auricotemporalis. This opening is present in turtles and diapsids, where it is traversed by the ramus cutaneous recurrens (r. auricotemporalis) of the mandibular nerve (V3). The foramen is absent in captorhinids, but numerous tiny pits in this region have been interpreted as exits for a network of minute branches of this nerve (Heaton 1979). Two smaller foramina, of unknown function, are present near the anterior end of the surangular. Articular. The articular forms the posterior portion of the lower jaw. It is overlapped ventrally by the angular, laterally by the surangular and medially by the prearticular. Only the posterior end and dorsal surface of the articular are not covered by other elements. The dorsal surface is almost horizontal, facing only very slightly outwards. This area is greatly expanded transversely, and bears the area articularis mandibularis (a.a.m.). 324 PALAEONTOLOGY, VOLUME 40 text-fig. 8. Pareiasuchus nasicornis Haughton and Boonstra, 1929a. a, reconstruction of premaxillary tooth, in labial view, b, reconstruction of partially erupted mandibular tooth, in lingual view, c, recon- struction of the preserved portions of the hyoid apparatus in ventral view. Scale bar represents 10 mm. The a.a.m. is covered by the quadrate in this specimen; presumably, as in other pareiasaurs, it consists of two antero-posteriorly elongated cotyles separated by a low ridge. The articular projects only a very short distance behind the a.a.m., forming a very short, broad retroarticular process. The posterior end of the process is blunt and is notched in the centre. Prearticular. The prearticular is a sheet-like element which forms much of the medial surface of the posterior end of the lower jaw. It overlaps the splenial anteriorly and overlaps the articular posteriorly. Dorsally, it overlaps the coronoid and forms the medial margin of the adductor fossa. This section of the margin curls medially, forming a prominent horizontal lip. Ventrally, the prearticular overlaps the angular and forms the dorsal margin of the foramen intermandibularis caudalis (see Angular). Coronoid. The coronoid is a small bone mostly covered laterally by the dentary, and medially by the splenial and prearticular. It forms a very small portion of the anterior rim of the adductor fossa. The coronoid process is not accessible to preparation in BP1 1 /3653 and is either poorly preserved or similarly inaccessible in the other specimens of P. nasicornis: it has been reconstructed by comparison with other pareiasaurs (e.g. Deltavjatia vjatkensis , PIN 2212/6). Splenial. The splenial forms much of the medial and ventral surface of the anterior portion of the mandible. Dorsally, it overlaps the coronoid. It overlaps the dentary almost completely, leaving just a thin splint exposed immediately below the tooth row. Ventrally, the splenial overlaps the angular, curving under it to meet the dentary on the ventral surface of the mandible. Two flanges project posteriorly from the splenial. The ventral flange is longer and enters the anterior margin of the foramen intermandibularis caudalis (see Angular). The dorsal flange overlaps the coronoid. Anteriorly, the splenial is deeply notched. The base of this notch forms the medial wall of the foramen intermandibularis medius (see Dentary). Two flanges extend anteriorly, above and below this notch. The shorter, dorsal flange terminates well before the symphysis. The longer, ventral flange forms the ventral border of the sulcus cartilagmis meckelii (see Dentary) and broadly sutures with its partner, forming the ventral half of the mandibular symphysis. The ventral portion of the symphysis projects posteriorly, resulting in a median ventral flange in the ‘chin' area. Marginal dentition General (Text-fig. 2b-c). All the marginal teeth are preserved. However, as the jaws are tightly closed, the lingual surfaces of the premaxillary and maxillary teeth are inaccessible, whilst all surfaces of most of the dentary teeth are difficult to prepare. One partially erupted dentary tooth, however, is located low down on the dentary, and it has been possible to prepare most of the lingual surface of this tooth. Upper marginal dentition (Text-figs 3a, 8a). All the premaxillary and maxillary teeth are deeply thecodont. The teeth decrease in size posteriorly, such that the posteriormost are half the dimensions of the anteriormost. They are arranged in a single row, and closely spaced so that adjacent crowns are very slightly overlapping. The dentition thus forms a continuous shearing edge. The teeth are inflected medially, so that the crowns point not directly downwards, but also inwards towards the palate. LEE ET AL.\ PERMIAN PAREIASAUR 325 The premaxillary and maxillary teeth are all very similar in shape. Each crown is fan-shaped, being compressed labiolingually and expanded anteroposteriorly along the jaw margin beyond the limit of the root. Nine to eleven cusps are spaced evenly around the crown. The three central cusps are slightly larger than the cusps along the anterior and posterior edges. The crown is slightly recurved, so that the apex curves slightly inwards towards the palate. Lower marginal dentition. The only tooth exposed on the dentary (Text-figs 3b, 8b) differs in shape from all the teeth in the upper jaw. The crown is taller, there are more cusps (13), and they extend further down along the anterior and posterior margins of the crown. Also, the cusps on the anterior and posterior margins point slightly to the side of the tooth (i.e. anteriorly or posteriorly), rather than directly terminally. The crown is also not recurved lingually. Exactly the same situation occurs in Pareiasuchus peringueyi (BPI 1/4105) and Scutosaurus karpinskii (Lee 19946): the upper and lower teeth of at least some pareiasaurs, therefore, were different. DISCUSSION Anatomical considerations The above description of Pareiasuchus nasicornis substantially modifies previous interpretations of the cranial anatomy. Boonstra (1934a) suggested that Pareiasuchus nasicornis (along with P. peringueyi ) possessed a very depressed snout, a low braincase and supraoccipital pillar, and a quadrate that is much inclined anteroventrally. However, these all appear to be taphonomic artefacts, Boonstra having based his observations only on the holotype skull of P. nasicornis , SAM 3016. This skull has been compressed dorsoventrally, producing the flat snout, braincase, and supraoccipital. This crushing also displaced the dorsal ramus of the quadrate from its normal vertical position into an oblique position. Boonstra (1934a) also suggested that the maxilla of P. nasicornis was shallow, and that the maxilla-jugal contact was absent. However, none of the sutures is very clear on his specimen, whereas BPI 1 /3653 clearly shows that the maxilla has an anterodorsal expansion and contacts the jugal. The sutures in the illustrations of Boonstra ( 1934a), which are all dotted and thus hypothetical, are also partly inaccurate. Apart from the anomalies already noted, the supernumerary element was assumed incorrectly to be absent, the parietals and supratemporal are much too narrow, and the postparietal is depicted as rectangular instead of U-shaped. Because of lack of adequate descriptions, Gauthier et al. (19886) misinterpreted the distribution of many characters in pareiasaurs. They asserted that pareiasaurs lacked both a maxilla-jugal contact and an enlarged foramen palatinum posterius (which they homologized with the diapsid suborbital foramen), and possessed a convex occipital condyle, a vomer that is narrower anteriorly and broader posteriorly, a postparietal which is occipital, and premaxillary teeth that are smaller than maxillary teeth. The present study of Pareiasuchus nasicornis reveals that the maxilla-jugal contact and enlarged foramen palatinum posterius are both present, and that the occipital condyle is concave, the vomer is broader anteriorly, the postparietal is exposed on the skull roof, and the premaxillary teeth are the largest in the tooth row. All other adequately known pareiasaurs are identical to P. nasicornis with respect to these characters (Lee, pers. obs.). The supratemporal element in pareiasaurs has been identified usually as a ‘tabular’ (e.g. Haughton and Boonstra 1929a; Hartmann-Weinberg 1933; Boonstra 1934a; Bystrow 1957). However, its position on the skull roof, rather than on the occiput, and its lack of contact with the supraoccipital, suggest that it is a supratemporal. In diadectomorphs, and all basal amniotes which retain both a supratemporal and a tabular, the supratemporal has these relationships, whereas the tabular is occipital and contacts the supraoccipital. Furthermore, a phylogenetic analysis of parareptiles (Lee 1995) has revealed that the successive outgroups of pareiasaurs are turtles, lanthanosuchids and procolophonoids: these taxa all lack a tabular but possess a large supratemporal. Hence, both topological similarity and phylogenetic continuity (character congruence) indicate that the large temporal element in pareiasaurs is a supratemporal, not a tabular. The identity of the ‘supernumerary bone' between the supratemporal and the postparietal is also disputed. Wild (1985) called it a ‘tabular’, whilst Brink (1955), Walker (1973) and Maxwell ( 1991 ) PALAEONTOLOGY, VOLUME 40 326 PALAEONTOLOGY, VOLUME 40 text-fig. 9. The phylogenetic relationships of Pareia- suchus nasicornis to other well-known pareiasaurs. Characters diagnosing the labelled groupings are discussed in the text. The position of turtles in this scheme is uncertain: as discussed in Lee (1996), they are either nested within pareiasaurs (as part of Clade D), or are the sister group to pareiasaurs (and thus are the sister group to Clade A). interpreted it as a cervical osteoderm that had been incorporated into the cranium. There is a third possibility: that the bone is a neomorphic ossification. Critical developmental evidence is lacking; the homology of the supernumerary bone cannot be established unequivocally. It is not a persistent tabular, since the tabular has been lost in all the outgroups to pareiasaurs. Furthermore, the supernumerary bone lies on the skull table, medial to the supratemporal, in a very different position from the tabular of other basal amniotes. There is thus neither phylogenetic continuity nor topographical similarity to suggest that the supernumerary bone in pareiasaurs is a tabular (see above). However, it is difficult to choose between the two remaining possibilities. Many taxa (e.g. ceratopsians, ankylosaurs) incorporated osteoderms into the cranium, and in the dwarf pareiasaur Nanoparia luckhoffi , an osteoderm has clearly been incorporated into the skull between the squamosal and the supratemporal (Broom 1936; Brink 1955). In all these cases, the osteoderms overlie the original cranial elements and are very superficial. The supernumerary bone in pareiasaurs, however, appears to be integrated completely into the cranium, and the evidence for it being an osteoderm is less conclusive. Thus, the possibility remains of it being a neomorph - an entirely new ossification centre appearing between the postparietal and supratemporal. Intriguingly, in one skull of Scutosaurus (PIN 156/2), the supernumerary element appears to have fallen out from the back of the skull, which is otherwise complete. This suggests that it was less firmly attached than the other skull bones and supports (albeit weakly) the view that it represents a modified osteoderm. The linear dimensions of BPI 1/3653 are about three-quarters of those of the largest specimens of P. nasicornis: 3016 (the type), GSP 475 and GSP CBT4. BPI 1 / 3653 also differs from the latter skulls in some other features, all related to cranial ornament. The angular boss is a low rounded swelling, rather than a horn-like process. The bosses around the quadratojugal are also less prominent. The maxilla is smooth, rather than ornamented. In these respects, it is similar to other small skulls here assigned to P. nasicornis: SAM K6607 and GSP TN257. These differences are almost certainly ontogenetic: cranial ornament is more prominent in larger individuals of many taxa (Dodson 1975). In particular, the same differences are found in specimens which have been interpreted as juvenile and adult Pareiasuchus peringueyi and Scutosaurus karpinskii (Lee 19946). The lack of a bony separation between the fenestra ovalis and the foramen jugular anterius might also be the result of incomplete ossification in an immature animal. The open sutures in BPI 1/3653 might also be a juvenile feature. Some sutures are also faintly visible on one of the other small skulls (SAM K6607) but are not visible on the other (GSP TN257), or on any of the large skulls. Sutures LEE ET AL.\ PERMIAN PAREI ASAUR 327 are also clearer in smaller individuals of Bradysaurus (Broom 1924) and Scutosaurus (Lee 19946). Finally, the fragmentary postcranial elements of this animal show one feature characteristic of juvenile reptiles: the neural arches and centra are separate (Kemp 1986). Phylogenetic implications The position o/ Pareiasuchus nasicornis within pareiasaurs. The information derived from this new specimen was included in a phylogenetic analysis which considered 128 osteological characters and all valid species of pareiasaurs, together with turtles, procolophonoids, lanthanosuchids, nyctiphruretids and nycteroleterids (Lee 19946). A summary of the results is presented in Text- figure 9, which shows the systematic position of P. nasicornis relative to the better-known taxa of pareiasaurs. As discussed in the diagnosis of the genus Pareiasuchus , Pareiasuchus nasicornis and Pareiasuchus peringueyi are immediate relatives, based on the following synapomorphies: the narrow, distally projecting ent- and ectepicondyles; the greatly anterodorsally inclined iliac shaft; the flattened second and third sacral ribs; and the sharply bent femoral shaft. Pareiasuchus is related to other pareiasaurs which possess heavy cranial ornament, i.e. Scutosaurus , Pareiasaurus , Elginia and Parasaurus. The dwarf pareisaurs ( Anthodon ) and the large, massively ossified forms (Bradysaurus) are very distantly related to this clade. Turtles might be part of the clade of dwarf pareiasaurs: a full discussion of this is presented elsewhere (Lee 19946). The characters diagnosing the groupings in Text-figure 9 are listed below: they are the unambiguous (uniquely derived and unreversed) characters identified in the analysis just mentioned (Lee 19946). For each character listed below, the derived state is listed first, while the primitive state (found in pareiasaurs outside the relevant clade, and in the nearest outgroups to pareiasaurs: lanthanosuchids, procolophonoids, nyctiphruretids, and nycteroleterids) follows in parentheses. These characters are discussed in detail and illustrated in Lee (19946). Clade A. All pareiasaurians. See Lee (1995). Clade B. (1) humeral torsion 45° or less (torsion 60°); (2) osteoderms with a distinct central boss (osteoderms smooth); (3) basal tubera positioned anteriorly, mid-way between basipterygoid processes and occipital condyle (basal tubera positioned closer to occipital condyle); (4) upper and lower marginal teeth with nine or more cusps (upper and lower marginal teeth with seven cusps); (5) cusps evenly spaced around tooth crown (cusps not evenly spaced - central or terminal three cusps much more close together than the cusps on the anterior and posterior margins); (6) 19 or fewer presacral vertebrae (20 or more presacrals); (7) cleithrum absent (cleithrum present in Bradysaurus and most of the outgroups, but also absent in derived procolophonoids). Clade C. (1) dentary teeth with 12 or more cusps (dentary teeth with 1 1 or fewer cusps); (2) 20 or fewer caudal vertebrae (more than 20 caudals); (3) dermal armour covers entire dorsum (dermal armour restricted to dorsal midline in Bradysaurus and Deltavjatia , and absent in the outgroups); (4) limbs covered in conical bony studs (limbs without dermal armour). Clade D. (1) scapula blade cylindrical (scapula blade flat and plate-like); (2) ulnar trochlea on humerus is a discrete tubercle (ulnar trochlear on humerus is a groove); (3) olecranon process reduced (olecranon process prominent in all other pareiasaurs, and in most related taxa, but convergently reduced in some procolophonoids); (4) trochanter major enlarged into a prominent tubercle (trochanter major present as a weak swelling in other pareiasaurs, absent in the other taxa); (5) trochanter major positioned very proximally, directly under capitelluin (trochanter major positioned distal to capitelluin in other pareiasaurs, absent in other taxa); (6) cnemial crest on tibia greatly reduced (cnemial crest is a prominent ridge); (7) osteoderms united over entire body (osteoderms separate). Clade E. (1) lingual surface of dentary teeth with a distinct triangular ridge, widest near the base and tapering distally (lingual surface of dentary teeth smooth); (2) anterior expansion of blade of ilium greatly everted, being almost horizontal (blade of ilium slightly everted, or not at all); (3) osteoderms with highly irregular, rugose ridges radiating from the central boss (osteoderms without irregular, rugose ridges). 328 PALAEONTOLOGY, VOLUME 40 Clade F. (1) tubercle on ventral surface of basipterygoid process (tubercle absent); (2) horn on maxilla, immediately posterior to external naris (horn absent). Clade G. Pareiasuchus. See generic diagnosis. Clade H. Pareiasuchus peringueyi. (1) Angular boss blunt (angular boss horn-like); (2) palatal ramus of the quadrate with large, posteriorly directed spine (spine absent). Clade I. Pareiasuchus nasicornis. See specific diagnosis. The position of pareiasaurs within Amniota. It has been proposed recently that pareiasaurs are the nearest relatives of turtles (Lee 1993, 1995). Alternative hypotheses are that the nearest relatives of turtles are captorhinids (Gaffney and Meylan 1988; Gauthier et al. 1988n, 19886), or procolophonoids (Reisz and Laurin 1991 ; Laurin and Reisz 1993, 1995). Gardiner’s (1982) radical proposal that dicynodonts are the nearest relatives of turtles has not found any acceptance (e.g. Gay 1987; King 1988; Lee 1995), and Gardiner himself no longer holds this view (see Gardiner 1993). A fuller treatment of turtle relationships is presented elsewhere (Lee 1995). However, it should be emphasized that all these discussions have been hampered by a lack of accurate, detailed descriptions of pareiasaur anatomy. The present excellent specimen sheds light on this problem in clearly showing several proposed pareiasaur-turtle synapomorphies (some of which have been inadequately described in pareiasaurs until now). It also demonstrates the presence in pareiasaurs of traits previously asserted to be restricted either to procolophonoids and turtles, or to captorhinids and turtles. This new information thus strengthens arguments for a pareiasaur-turtle clade and weakens alternative suggestions. Pareiasuchus nasicornis possesses the following traits, proposed to be pareiasaur-turtle synapomorphies (Lee 1995): a large foramen palatinum posterius; a medially positioned choana; a blunt cultriforin process; a horizontal transverse flange of the pterygoid; opisthotic-squamosal suture; a long lateral flange of the exoccipital; and a narrow supraoccipital with a long sagittal suture with the skull table. These are present in other adequately known pareiasaurs (Lee, pers. obs.). The described specimen also casts doubt over many proposed procolophonoid-turtle synapo- morphies. BPI 1 /3653 shows that Pareiasuchus nasicornis , like other adequately known pareiasaurs, has a short cultriform process, a wide antorbital buttress formed by the prefrontal and palatine, an anterior expansion of the maxilla, an occipital flange of the squamosal, an enlarged quadratojugal, a notched anterior end of the splenial, and a medially enclosed adductor fossa. These traits have recently been proposed as procolophonoid-turtle synapomorphies (Reisz and Laurin 1991 ; Laurin and Reisz 1993, 1995). Discovery of them in pareiasaurs indicates that they diagnose a more inclusive grouping. A further proposed procolophonoid-turtle synapomorphy, the large basal tubera on the basioccipital, is not found in primitive turtles such as Proganochelys. Furthermore, as shown in the above description, P. nasicornis , like all pareiasaurs, also has basal tubera. These are formed mainly by the basisphenoid, but their posterior portions are formed by the basioccipital. The above description also has relevance to the proposed captorhinid-turtle clade. BPI 1/3653, like all pareiasaurs (Lee, pers. obs.), has a foramen-orbitonasale, and lacks a tabular: these traits have previously been interpreted as the most compelling captorhinid-turtle synapomorphies (Gaffney and Meylan 1988). Lack of adequate published information on pareiasaur anatomy has caused previous workers to misinterpret the distribution of these characters. Some previous descriptions of pareiasaurs have suggested incorrectly that, with respect to the above characters, some pareiasaurs differ from P. nasicornis. Flaughton and Boonstra (19296) depicted a tapering, rather than forked, splenial in some pareiasaurs (e.g. Anthodon serrarius , Propappus omocratus). However, examination of the relevant specimens reveals that in each case the dorsal prong of the fork in the splenial has been lost through crude preparation. All adequately prepared pareiasaur material shows the forked splenial (Lee, pers. obs.). Ivachnenko (1987) has depicted some individuals of Scutosaurus with a long cultriform process: examination of the material of this taxon, including the specimens used in his reconstructions, reveals the cultriform process to be short. Similarly, a few illustrations of Proganochelys in Gaffney (1990) depict a long LEE ET AL.. PERMIAN PAREIASAUR 329 cultriform process; however, most depict a short, blunt process, which is the actual condition (Gaffney, pers. comm. 1995). Thus, the preceding description of the excellently preserved skull helps confirm many pareiasaur- turtle synapomorphies, and simultaneously casts doubt over many of the synapomorphies proposed for alternative groupings. The problem of whether turtles are related to procolophonoids or to pareiasaurs is discussed more fully elsewhere (Lee 1994 b, 1995) and it was found that only two of the 17 procolophonoid-turtle synapomorphies recently proposed by Laurin and Reisz ( 1995) appear to be valid, while over 20 derived characters unite pareiasaurs and turtles. FUNCTIONAL MORPHOLOGY The skull of Pareiasuchus nasicornis is heavily ossified, solid, and completely akinetic: the braincase is sutured to the skull roof via the supraoccipital, the paroccipital process, and presumably the pleurosphenoid. The palate is held firmly in place, not only by the usual contacts with the quadrate and above the alveolar ridge, but also via the fused basicranial articulation and the robust antorbital buttress. The pareiasaur skull, as exemplified by P. nasicornis , clearly reflects herbivorous habits, despite two previous suggestions to the contrary. Hartmann-Weinberg (1937) suggested that pareiasaurs were aquatic predators that fed on amphibian larvae, based largely on the co-occurrence of the two taxa in the same deposits in Russia. Case (1926) suggested that they consumed floating algal mats, based on their 'weak' dentition. However, these interpretations of pareiasaurs as aquatic feeders are not supported by their postcranial morphology, which exhibit no indications of aquatic habits (e.g. Lee 1994u). Conversely, pareiasaurs possessed many other features well-known to be characteristic of terrestrial herbivores. The 'weak’ dentition is, in fact, quite typical of herbivores. The teeth are labiolingually compressed and expanded anteroposteriorly, bearing numerous large cusps arranged in a single row around the edge. Such teeth are characteristic of herbivorous terrestrial reptiles, such as numerous groups of ornithischian dinosaurs (e.g. stegosaurs, fabrosaurids), synapsids (e.g. caseids) and squamates (e.g. iguanids). In extant forms, they have been shown to be correlated with herbivory (Hotton 1955; Montanucci 1968). The deep tooth sockets in pareiasaurs allowed replacement teeth to reach a large size before the old tooth was shed, allowing rapid replacement of teeth and thus, reducing the number of gaps in the cutting edge (Edmund 1969; Throckmorton 1976). The anteroposteriorly expanded teeth, lack of interdental spaces, and isodont dention combine to form an uninterrupted, even cutting edge, very different from the puncturing dentition typical of most basal amniotes, which were insectivorous (Carroll 1964, 1982). Such a cutting edge is characteristic of herbivores that employ a cropping action (Montanucci 1968; Norman 1984; Galton 1986). This cropping action does not require contact between the upper and lower marginal teeth: rather, the upper and lower tooth rows slide closely past one another in a scissor-like action (Throckmorton 1976). Accordingly, the teeth of pareiasaurs never show any wear facets or other signs of attrition (tooth-to-tooth wear; Osborn and Lumsden 1978). The reduction of the transverse flange (see above) is also correlated with herbivory. This structure represents the origin of the anterior pterygoideus muscle (part of the internal adductors). This muscle, which pulls upwards and forwards, acts most effectively (perpendicular to the long axis of the lower jaw) when the jaws are widely open, and is most important in the kinetic-inertial feeding systems of carnivores (e.g. Olson 1961). It also serves to guide the lower jaw and helps prevent dislocation when the jaws are closed against struggling prey. In reptilian herbivores, which have a static feeding system (Olson 1961), the anterior pterygoideus is less important, and problems with struggling prey do not occur, and thus the transverse flange of the pterygoid is reduced (e.g. King el a/. 1989; King 1990). The coronoid process is also reduced, compared with the condition in most other 'parareptiles’ such as nyctiphruretids, procolophonoids and turtles. The medial external adductors which insert on the coronoid process pull backwards and upwards on the process, and thus function most effectively 330 PALAEONTOLOGY, VOLUME 40 when the coronoid process is oriented perpendicular to the line of muscle action (i.e. when the jaws are wide open). The coronoid process, like the transverse flange of the pterygoid, is emphasized in the kinetic-intertial jaws of carnivores, and is reduced in herbivores (Janis 1995). The pareiasaurian mandible is very heavy. The anterior position of the jaw articulation means that the lower jaw of pareiasaurs is rather short. Given the weakness of the pterygoideus complex, the heaviness of the mandible, and the shortness of the lower jaw, the gape of pareiasaurs was undoubtedly rather narrow, and the jaw movements slow. All these features are consistent with herbivory. As in many herbivores, the jaw articulation is positioned well below the tooth row (Norman 1984; Galton 1986). This arrangement allows for simultaneous occlusion of cheek teeth (Greaves 1995; Janis 1995). Also, Davis (1964) has demonstrated that, when the jaw articulation is raised above the level of the tooth row, an anteriorly directed shearing force is generated during occlusion, even if the muscles act perpendicular to the long axis of the lower jaw (Text-fig. 10a). Using the same model, it can be shown that if the articulation is positioned below the tooth row, as in Scutosaurus , a posteriorly directed force is generated during occlusion (Text-fig. 10b). The posterior forces are most pronounced near the rear of the tooth row, for geometric reasons (Text-fig. 10b). Such shearing forces have been demonstrated to facilitate cropping of tough plant matter (Throckmorton 1976). Although pareiasaurs were clearly herbivores, there are no indications of cheeks, no complex tooth batteries or other elaborate masticatory adaptations. Oral processing of food appears to have been limited to cropping bits of vegetation before swallowing. Such feeding behaviour is characteristic of extant iguanid squainates (Throckmorton 1976), which have dentition very similar to that of pareiasaurs. In pareiasaurs, the cropping action would have been facilitated by propalinal movements. Given their large size and broad snouts (see Jarman 1974; van Soest 1982; Janis and Erhardt 1988), pareiasaurs were probably not very selective feeders, ingesting large amounts of low- quality food. There is no evidence of gastralia, despite the fact that pareiasaur skeletons are commonly found articulated (Kitching 1977). Pareiasaurs therefore needed to process large quantities of low-grade food, but lacked the efficient equipment (dentition, gastralia) to break it down mechanically. They presumably coped by storing large amounts in the fermentation chambers for long periods, resulting in their large, bulky bodies. The function of the distinctive cheek flanges of pareiasaurs has never been properly discussed. Among extant taxa, edentate mammals have a projection in this area, which serves as an origin for muscles which move the lower jaws laterally during mastication (Naples 1982). However, a transversely oriented power stroke is known only in mammals and certain ornithopod dinosaurs, and in each case is accompanied by complex jaw adaptations (Weishampel 1984). The saddle- shaped jaw articulation, and shearing dentition of pareiasaurs obviously combined to preclude lateral grinding jaw movements (see King et al. 1989 for similar arguments pertaining to dicynodonts). Hence, the possibility that the pareiasaur cheek flanges served a similar function can be discounted. Pareiasaurs appear to have been very slow-moving animals, were too large to remain cryptic, and possessed no obvious offensive weapons. They appear to have adopted passive anti- predator defences: large size, blunt protective bosses over delicate organs such as the eyes and nostrils, and dermal armour over the body and limbs (Gow 1977; Lee 1994n). Interpreting the cheeks as defensive structures is therefore consistent with information derived from other areas of the body. The rows of large bosses on the edge of these cheek flanges are also consistent with this interpretation. Nopsca (1928) long ago remarked that the cheeks of pareiasaurs are very similar to those of ankylosaurs. Ankylosaurs and pareiasaurs both possessed strong defensive structures in other parts of the body, and the cheek flanges in these taxa may have helped protect the vulnerable throat region. The role of the ventrally directed boss on the mandible also remains uncertain. It is clearly partly an allometric feature. It is a small, rounded boss in young pareiasaurs (such as this individual of Pareiasuchus nasicornis) and dwarf forms (such as Nanoparia pricei BPI 1/7), but is a prominent conical spike in all adults of large pareiasaurs. Similarly, very large individuals of the procolophonoid Procolophon possess a weak rugosity in this area, but this feature is absent in LEE ET AL.\ PERMIAN PAREI ASAUR 331 text-fig. 10. Illustration of how simple elevation or depression of the jaw articulation results in shearing movements during occlusion, even in the absence of anteroposterior sliding (propalinal) movements at the jaw articulation. Arrow shafts are concentric to the jaw articulation, and represent the trajectory of the lower jaw during occlusion, a, an elevatedjaw articulation results in a forward shearing of the lower jaw during occlusion. b, a depressed jaw articulation, as found in Scutosaurus and other pareiasaurs, results in a backwards shearing of the lower jaw during occlusion. smaller individuals. Although obviously partly a correlate of large size, the observation that the angular boss was retained in all lineages of large pareiasaurs suggests some functional significance, but no clear explanation is apparent. Perhaps, like the cheek flanges, it helped to protect the vulnerable throat region. The only other amniotes with a boss in this region are entelodont mammals. However, the only comprehensive study of their functional cranial anatomy (Joeckel 1990) did not speculate on the function of the mandibular boss. Furthermore, the rest of the cranial architecture in entelodonts is so different from that of pareiasaurs that using these animals as analogues for one another would be very unwise. Acknowledgements. ML gratefully acknowledges financial assistance from The Association of Commonwealth Universities, The British Council, The Cambridge Philosophical Society, The Balfour Fund, Queens’ College (Cambridge), and the Australian Research Council, and thanks Rick Shine and colleagues for use of laboratory and office facilities and Malcom Ricketts for photographic assistance during the final stages of preparation of this manuscript. We are indebted to the following people for hospitality and access to specimens during museum visits: Gillian King, Madel Joubert, Clive Booth and Roger Smith (South African Museum), Oleg Lebedev, Michael Ivachnenko and Nick Kalandadze (Palaeontological Institute, Moscow), and Andre Keyser and Francois Durand (Geological Survey, Pretoria). REFERENCES araujo, d. c. f. 1984. Sistematica e taxonomia dos pareiasaurios : historico e perspectivas atuais. Pesquisas , 16, 227-249. — 1989. Estudo do material Sul-Americano de Pareiasauroidea. VI. Descriyao do esqueleto axial de Pareiasaurus americanus Araujo, 1985. Anais do Academie Brasiliera Ciencas , 61, 295-309. boonstra, l. d. 1929a. Pareiasaurian studies. Part III. On the pareiasaurian manus. Annals of the South African Museum , 28, 97-112. — 19296. Pareiasaurian studies. Part IV. On the pareiasaurian pes. Annals of the South African Museum , 28, 113-122. — 1932. Pareiasaurian studies. Part VIII. The osteology and myology of the locomotor apparatus. Part B. Fore limb. Annals of the South African Museum , 28, 436-503. — 1934a. Pareiasaurian studies. Part IX. The cranial osteology. Annals of the South African Museum , 31, 1-38. — 19346. Pareiasaurian studies. Part X. The dermal armour. Annals of the South African Museum , 31, 39-48. 332 PALAEONTOLOGY, VOLUME 40 - 1934c. Pareiasaurian studies. Part XI. The vertebral column and ribs. Annals of the South African Museum , 31, 49-66. brink, a. s. 1955. On Nanoparia. Palaeontologia Africana , 3, 57-63. broom, R. 1909. An attempt to determine the horizons of the fossil vertebrates of the Karroo. Annals of the South African Museum , 7, 285-289. - 1924. On some points in the structure of the pareiasaurian skull. Proceedings of the Zoological Society of London, 1924. 499-508. 1936. On some new genera and species of Karroo fossil reptiles, with notes on some others. Annals of the Transvaal Museum , 18, 349-386. 1940. On some new genera and species of fossil reptiles from the Karroo beds of Graaf-Reinet. Annals of the Transvaal Museum , 20, 157-192. and haughton, s. h. 1913. On the skeleton of a new pareiasauran ( Pareiasuchus peringueyi, g. et. sp. nov.). Annals of the South A frican Museum, 12, 17-25. bryant, h. n. and russell, a. p. 1992. The role of phylogenetic analysis in the inference of unpreserved attributes of extinct taxa. Philosophical Transactions of the Royal Society of London, Series B, 337, 405^118. bystrow, a. p. 1957. [The pareiasaur skull.] Trudy Paleontologicheskogo Instituta, Akademiia Nauk SSSR, 68, 3-18. [In Russian]. Carroll, r. l. 1964. The earliest reptiles. Zoological Journal of the Linnean Society, 45, 61-83. 1982. The early evolution of reptiles. Annual Review of Ecology and Svstematics, 13, 87-109. case, E. c. 1926. Environment of tetrapod life in the late Paleozoic of regions other than North America. Publication of the Carnegie Institute of Washington, 375, 1—211. davis, d. d. 1964. The giant panda: a morphological study of evolutionary mechanisms. Fieldiana , Zoology Memoir , 3, 1-339. dodson, p. 1975. Taxonomic implications of relative growth in lambeosaurine hadrosaurs. Systematic Zoology, 24, 37-54. drysdall, a. r. and KiTCHiNG. J. w. 1963. A re-examination of the Karroo succession and fossil localities of part of the Upper Luangwa Valley. Memoir of the Geological Survey of Northern Rhodesia , 1, 1-62. edmund, a. G. 1969. Dentition. I 17-200. In gans, c., bellairs, a. d’A., parsons, t. s. (eds). Biology of the Reptilia. Volume 1. Academic Press, London, 373 pp. Gaffney, e. s. 1979. Comparative cranial morphology of Recent and fossil turtles. Bulletin of the American Museum of Natural History, 164, 65-376. - 1990. The comparative osteology of the Triassic turtle Proganochelys. Bulletin of the American Museum of Natural History, 194. 1-263. and meyland, p. a. 1988. A phylogeny of turtles. 157-219. In benton, m. j. (ed. ). The phytogeny and classification of tetrapods. Clarendon Press, Oxford, 377 pp. galton, p. m. 1986. Herbivorous adaptations of Late Triassic and Early Jurassic dinosaurs. 203-221. In padian, K. (ed.). The beginning of the age of dinosaurs. Cambridge University Press, Cambridge, 378 pp. gardiner, b. g. 1982. Tetrapod classification. Zoological Society of the Linnean Society, 74, 207-232. - 1993. Haematothermia: warm-blooded amniotes. Cladistics, 9, 368-395. Gauthier, j. a., kluge, a. G. and rowe, t. 1988tf. Amniote phylogeny and the importance of fossils. Cladistics, 4, 105-209. 19887). The early evolution of the Amniota. 103-155. In benton, m. j. (ed ). The phylogeny and classification of the tetrapods. Clarendon Press, Oxford, 377 pp. gay, s. a. 1987. A dicynodont fauna from the Permian of Tanzania. Unpublished Ph.D dissertation. University of Cambridge, U.K. gow, c. E. 1977. Owenetta in perspective. Palaeontologia Africana , 20, 115-1 18. greaves, w. s. 1995. Functional predictions from theoretical models of the skull and jaws in reptiles and mammals. 99-115. In thomason, j. j. (ed.). Functional morphology in vertebrate paleontology. Cambridge University Press, Cambridge, 277 pp. haekel, e. 1866. Generelle Morphologie der Organismen. Reimes, Berlin. hartmann-weinberg, a. 1933. Die Evolution der Pareiasauriden. Trudy Paleontologicheskogo Instituta, Akademiia Nauk SSSR, 3, 3-66. 1937. Pareiasauriden als leitfossilien. Problemy paleontologii, 2/3, 649-712. haughton, s. H. and boonstra, l. d. 1929u. Pareiasaurian studies. Part I. An attempt at a classification of the Pareiasauria based on skull features. Anna/s of the South A frican Museum, 28, 79-87. 1929/). Pareiasaurian studies. Part V. On the pareiasaurian mandible. Annals of the South African Museum, 28, 261 289. LEE ET AL.\ PERMIAN PAREIASAUR 333 — 1930. Pareiasaurian studies. Part VI. The osteology and mycology of the locomotor apparatus. A. Hind limb. Annals of the South African Museum , 28, 297-367. heaton, m. j. 1979. The cranial anatomy of primitive captorhinid reptiles from the Pennsylvanian and Permian of Oklahoma and Texas. Bulletin of the Oklahoma Geological Survey , 127, 1-84. hotton, n., hi. 1955. A survey of adaptive relationships of dentition to diet in the North American Iguanidae. American Midland Naturalist, 53, 88-114. huene, f. von 1944. Pareiasaurierreste aus dem Ruhu-Gebiet. Palaontologische Zeitschrift , 23C, 386-410. ivachnenko, m. f. 1987. [Permian parareptiles of the USSR.] Trudy Paleontologicheskogo lnstituta, Akademiia Nauk SSSR , 223, 3-16. [In Russian], janis c. m. 1995. Correlations between craniodental morphology and feeding behavior in ungulates: reciprocal illumination between living and fossil taxa. 76-98. In thomason, j. j. (ed. ). Functioned morphology in vertebrate paleontology. Cambridge University Press, Cambridge, 277 pp. — and erhardt, d. 1988. Correlation of relative muzzle width and relative incisor width with dietary preference in ungulates. Zoological Journal of the Linnean Society , 92, 267-284. jarman, p. J. 1974. The social organisation of antelope in relation to their ecology. Behaviour, 58, 215-267. joeckel, R. m. 1990. A functional interpretation of the masticatory system and paleoecology of entelodonts. Paleobiology , 16, 459^482. kemp, T. s. 1986. The skeleton of a baurid therocephalian therapsid from the Lower Triassic (Lystrosaurus zone) of South Africa. Journal of Vertebrate Paleontology, 6, 215-232. keyser, a. w. and smith, r. m. h. 1978. Vertebrate biozonation of the Beaufort Group with special reference to the western Karoo Basin. Annals of the Geological Survey of South A frica, 12, 1-35. king, G. M. 1988. Anomodontia. Handbuch der Palaoherpetologie. Ted 17C. Gustav Fischer Verlag, Stuttgart, 174 pp. - 1990. The dicynodonts : a study in palaeobiology. Chapman and Hall, London, 233 pp. — oelofsen, b. w., and rubidge, b. s. 1989. The evolution of the dicynodont feeding system. Zoological Journal of the Linnean Society, 96, 185-211. kitching, J. w. 1963. The fossil localities and mammal-like reptiles of the Upper Luangwa Valley, Northern Rhodesia. South African Journal of Science, 59, 259-264. — 1970. A short review of the Beaufort zoning in South Africa. 309-312. In haughton, s. h. (ed.). I.U.G.S. Second Symposium on Gondwana Stratigraphy and Palaeontology. C.S.I.R., Pretoria, 689 pp. — and 1977. Distribution of the Karroo vertebrate fauna. Memoir of the Bernard Price Institute for Palaeontological Research , University of Witwatersrand (Johannesburg), 1, 1—131. kuhn, o. 1969. Cotylosauria. Handbuch der Palaoherpetologie. Ted 6. Gustav Fischer Verlag, Stuttgart, 89 pp. laurenti, j. n. 1768. Specimen medicum, exhibens synopsin Reptilium emendation, cum experimentalis circa Venena. Vienna, 217 pp. laurin, m. and reisz, r. r. 1993. The origin of turtles. Journal of Vertebrate Paleontology, 13, 46A. - 1995. A reevaluation of early amniote phylogeny. Zoological Journal of the Linnean Society, 113, 165-223. lee, m. s. y. 1993. The origin of the turtle body plan: bridging a famous morphological gap. Science, 261, 1716-1720. 1994 a. The turtle's long lost relatives. Natural History, 103 (6), 63-65. — 19946. Evolutionary morphology of pareiasaurs. Unpublished Ph.D Dissertation, University of Cambridge, U.K. — 1995. Historical burden in systematics and the interrelationships of parareptiles. Biological Reviews , 70, 459-547. — in press. A taxonomic revision of pareiasaurs: implications for Permian terrestrial palaeoecology. Modern Geology. maxwell, w. d. 1991 . The pareiasaur Elginia from Elgin, north-east Scotland, and the Late Permian extinction event. Unpublished Ph.D Thesis, Queen’s University, Belfast. montanucci, r. r. 1968. Comparative dentition in four iguanid lizards. Herpetologica, 24, 305-315. Naples, v. l. 1982. Cranial osteology and function in tree sloths, Bradypus and Choloepus. American Museum Novitates, 2739. 1—41. newton, e. t. 1893. On some new reptiles from the Elgin sandstones. Philosophical Transactions of the Royal Society of London, Series B, 184, 473 — 489. nopcsa, F. 1928. Palaeontological notes on reptiles. Geologica Hungarica, 1, 1-84. norman, d. b. 1984. On the cranial morphology and evolution of ornithopod dinosaurs. 521-547. In ferguson, m. w. (ed.). The structure , development and evolution of reptiles. Academic Press, London, 697 pp. 334 PALAEONTOLOGY, VOLUME 40 olson, E. c. 1961. Jaw mechanisms: rhipidistians, amphibians, reptiles. American Zoologist , 1, 205-215. osborn, j. w. and lumsden, a. G. s. 1978. An alternative to ‘thegosis’ and a re-examination of the ways in which mammalian molars work. Neites Jahrbuch fur Geologie und Paldontologie, Abhandlungen , 156, 371-392. Owen, r. 1876. Catalogue of the fossil Reptilia of South Africa. British Museum, London, 88 pp. queiroz, K. de and Gauthier, j. a. 1992. Phylogenetic taxonomy. Annual Review of Ecology and Systematics, 23, 449-480. reisz, r. r. and laurin, m. 1991. Owenetta and the origin of turtles. Nature. 349, 324-326. Seeley, h. G. 1888. Researches on the structure, organisation, and classification of the fossil Reptilia. II. On Pareiasaurus bombidens (Owen), and the significance of its affinities to amphibians, reptiles and mammals. Philosophical Transactions of the Royal Society of London , 179, 59-109. soest, p. j. van 1982. Nutritional ecology of the ruminant. O and B Books, Corvalis, Oregon, 476 pp. Throckmorton, G. s. 1976. Oral food processing in two herbivorous lizards. Iguana iguana (Iguanidae) and Uromastix aegypticus (Agamidae). Journal of Morphology, 148, 363-390. walker, a. d. 1973. The age of the Cuttie's Hillock Sandstone (Permo-Triassic) of the Elgin area. Scottish Journal of Geology, 9, 177-183. weishampel, D. B. 1984. Evolution of jaw mechanisms in ornithopod dinosaurs. Advances in Anatomy, Embryology and Cell Biology, 87, 1-106. wild, r. 1985. Ein Schadelrest von Parasaurus geinitzi H. v. Meyer (Reptilia; Cotylosauria) aus dem Kupferschiefer (Permian) von Richelsdorf (Hessen). Geologische Blatter fur Nordost-Bayern ( Gedenkschrift Freyberg), 34/35, 897-920. witmer, l. m. 1995. The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils. 19-33. In thomasen, J. j. (ed.). Functional morphology in vertebrate palaeontology. Cambridge University Press, Cambridge, 277 pp. m. s. y. lee Zoology Building A08 School of Biological Sciences University of Sydney, NSW 2006 Australia c. E. GOW j. w. kitching Bernard Price Institute for Palaeontological Research University of Witwatersrand Typescript received 1 1 December 1995 Private Bag 3, P. O. Wits 2050 Revised typescript received 12 August 1996 Johannesburg, South Africa APPENDIX: ANATOMICAL ABBREVIATIONS af adductor fossa mt median tubercle afor anterior foramen na nasal ang angular oc occipital condyle art articular op opisthotic bo basioccipital OSS ossification over naris bpt basipterygoid process pap paroccipital process bt basal tubera par parietal bs basisphenoid pal palatine cerll second ceratobranchial pfor posterior foramen ch choana pob postorbital cone concavity pof postfrontal cop copula PP postparietal cor coronoid pra prearticular cp cultriform process pro prootic era crista alaris prcl processus clinoideus ers crista supraoccipitalis prf prefrontal den dentary prm premaxilla LEE ET AL.\ PERMIAN PAREIASAUR 335 ec ectopterygoid Pt pterygoid ex exoccipital qa quadrate fcb foramen caroticum basisphenoidal qj quadratojugal fic foramen intermandibularis caudalis rp retroarticular process fja foramen jugular anterius scm sulcus cartilaginous meckelli fdm foramen dentofaciale majus sn supernumerary fim foramen intermandibularis medius so supraoccipital fm foramen magnum spl splenial fna foramen nervi auricotemporalis sq squamosal fo fenestra ovalis st supratemporal fpp foramen palatinum posterius sur surangular fr frontal VO vomer ju jugal V trigeminal foramen la lacrimal VII facial foramen max maxilla EARLY CAMBRIAN BRACHIOPODS FROM NORTH GREENLAND by LEONID POPOV, LARS E. HOLMER, ALBERT J. ROWELL and JOHN S. PEEL Abstract. A silicified late Early Cambrian (early Toyonian) brachiopod assemblage is described from the Paralleldal Formation of Peary Land, central North Greenland. The fauna comprises Acareorthis profunda sp. nov., Pelmanella borealis gen. et sp. nov., Kutorgina cingulata , and Agyrekia aff. obtusa. The assemblage is comparable to faunas described from the Lower Cambrian (Botomian) of south Tien-Shan, Kirgizia, and the lower Middle Cambrian (Ordian) of New South Wales, Australia. Primitive articulatory structures are described for the first time in Kutorgina cingulata , suggesting a close affinity to the Nisusiidae and supporting the assignment of the latter family to the Kutorginida. Acareorthis is reported for the first time outside Australia; it is here referred to the Chileida in view of its strophic shell lacking all traces of articulatory structures. The new obolellid family Pelmanellidae is erected, comprising Pelmanella and the Australian genus Bynguanoia. Our knowledge of late Early to early Mid Cambrian calcareous-shelled brachiopods is based mainly on a few silicified faunas from New South Wales, Australia (Roberts and Jell 1990), south Tien-Shan, Kirgizia (Popov and Tikhonov 1990), and Israel and Jordan (Cooper 1976). New collections of Early Cambrian silicified calcareous brachiopods described here from North Greenland contain a total of four taxa. Kutorgina cingulata (Billings, 1861) is the most abundant species and represents about 72 per cent, of the total number of individuals, whereas Agyrekia aff. obtusa Koneva, 1979 (5 per cent.), Acareorthis profunda sp. nov. (10 per cent.) and Pelmanella bdrealis gen. et sp. nov. (13 per cent.) are relatively rare. Finer details of morphology, such as micro-ornamentation and mantle canals, are not preserved in the available specimens, possibly due to the relatively coarse silicification of the shells. Elowever, other characters including the main internal morphology and articulation, which otherwise remain poorly known in the majority of Cambrian brachiopod taxa, are well preserved. In particular, the primitive articulation can be studied for the first time in the type species of Kutorgina , K. cingulata. Previously, kutorginid articulation was known only from Kutorgina catenata from the Botomian of Kirgizia (Popov and Tikhonov 1990). An unusual pattern of articulation is described in the new obolellid Pelmanella. GEOLOGICAL SETTING All Greenland material described herein was collected in 1980 from the lower part of the Paralleldal Formation (Higgins et al. 1991 ; Ineson and Peel in press) by J. S. Peel and P. D. Lane during the North Greenland Project (1978-80; 1984-85) organized by the Geological Survey of Greenland (Gronlands Geologiske Undersogelse). The collection (GGU 274907) was made on the north side of Paralleldal about 12 km north-east of the head of Jorgen Bronlund Fjord, southern Peary Land, central North Greenland (Text-fig. 1 ). At this locality, a diverse fauna includes archaeocyathids indicative of a late Early Cambrian (mid-late Toyonian) age (Debrenne and Peel 1986), olenellid trilobites, Salterella , the stenothecid Cambridium , helcionelloid molluscs ( Yochelcionella , La touch- el/a) and the brachiopods described below. Originally calcareous organisms are generally coarsely silicified. The Paralleldal Formation consists of about 140 m of cliff-forming dolomites which range in lithology from laminated, nodular dolomites (lower third) to oolitic and bioclastic dolomites; thick | Palaeontology, Vol. 40, Part 2, 1997, pp. 337-354, 1 pl.| © The Palaeontological Association 338 PALAEONTOLOGY, VOLUME 40 text-fig. 1 . Position of the collection locality within the Paralleldal Formation (Bronlund Fjord Group), north of Jorgen Bronlund Fjord (JBF), Peary Land, North Greenland. dolomite breccias are conspicuous in the middle and upper parts. The formation represents carbonate platform margin deposition within the southern shelf sequence of the Franklinian Basin succession of North Greenland and the Canadian Arctic Islands (Peel and Sonderholm 1991; Ineson and Peel in press). In the area north of Jorgen Bronlund Fjord, the Paralleldal Formation is the uppermost formation of the Bronlund Fjord Group; it is overlain unconformably by late early Ordovician dolomites of the Wandel Valley Formation (Ineson and Peel in press). GENERAL AFFINITIES OF THE FAUNA The described assemblage from the Paralleldal Formation includes genera of the Chileida, Obolellida (suborder Naukatidina) and Kutorginida, all of which became extinct at the end of the mid Cambrian. Surprisingly, no early orthides (Protorthidae or Eoorthidae) are present, although orthides are otherwise abundant in the late Early Cambrian brachiopod fauna of Israel (Cooper 1976) and the early Mid Cambrian (Ordian) fauna of New South Wales (Roberts and Jell 1990). Nisusia is also absent, but undescribed species of this genus were recovered in the underlying Saeterdal Formation, of late Botomian age. The Greenland assemblage is most similar to that from New South Wales described by Roberts and Jell (1990). Of the forms documented from Australia, Trematosia sp. cf. T. undulata Cooper is here referred to Kutorgina , while Kutorgina! sp. (Roberts and Jell 1990, p. 285) is probably comparable to Agyrekia aft', obtusa Koneva in the Greenland assemblage. The chileide Acareorthis is known only from New South Wales and Greenland. The presumably Botomian assemblage from Kirgizia (Popov and Tikhonov 1990) is also comparable; the cosmopolitian Kutorgina and Nisusia are the most abundant genera in Kirgizia, while the chileides (Chile) and naukatides ( Naukat , Oina) are represented only by endemic genera. Pelman (1977) described a late Early Cambrian (Botomian and Toyonian) assemblage from Siberia, characterized mainly by Kutorgina and Nisusia , but lacking naukatides and protorthoids. The obolellides ( Siberia , Alisina and Trematobolus) are also an important part of the Siberian fauna, but chileides ( Kotujella and Matutela) are rare, appearing in the upper Toyonian and occurring also in the Amgian (Pelman 1977). POPOV ET A L.. EARLY CAMBRIAN BRACHIOPODS 339 SYSTEMATIC PALAEONTOLOGY Abbreviations in figures and text are as follows: Cl, length of cardinal muscle field; Cw and Clw, width of cardinal muscle fields; II, length of pseudointerarea; Iw, width of pseudointerarea and width along posterior commissure; L, sagittal valve length; Lk, length of valve at commissure plane; max, maximum; min, minimum; N, number of specimens; Pw, width of dorsal median groove; S, standard deviation; SI, length of median ridge; T, height of valve; VP1, length of ventral muscle platform; VPw, width of ventral muscle platform; W, maximum width; X, mean. Dimensions are in millimetres. Figured and cited specimens from Greenland (prefixed by MGUH) are housed in the Geological Museum, University of Copenhagen. Other illustrated specimens are deposited in the Swedish Museum of Natural Elistory (RM Br) and Central Scientific Research Geologic Exploration Museum, St Petersburg (CNIGR). Order chileida Popov and Tikhonov, 1990 Superfamily matutelloidea Andreeva, 1962 Family chileidae Popov and Tikhonov, 1990 Genus acareorthis Roberts in Roberts and Jell, 1990 Type species. Acareorthis jelli Roberts (in Roberts and Jell), 1990; from the early Mid Cambrian (Ordian) First Discovery Limestone, New South Wales, Australia. Diagnosis. See Roberts and Jell 1990, p. 268. Remarks. Roberts (in Roberts and Jell 1990) regarded Acareorthis as a primitive orthide because it has a strophic, bisulcate shell, with costellate radial ornamentation; it was referred provisionally to the family Nisusiidae. Flowever, Acareorthis completely lacks articulatory structures with the exception of a weak transverse ridge along the straight dorsal posterior margin. Moreover, the perforation anterior to the ventral umbo suggests close affinity to the early Cambrian Chile (Popov and Tikhonov 1990). Acareorthis is referred here to the family Chileidae. It differs from Chile in having costate ornamentation covering the entire shell, a high ventral interarea with narrow, triangular delthyrium covered completely by a ridge-like pseudodeltidium, and in lacking a colleplax. Acareorthis is somewhat similar to Matutella (Cooper 1951; Andreeva 1962) in having a pseudodeltidium, radial ornamentation covering all the shell, and a ventral umbonal perforation. It differs from the latter genus, however, in possessing a planoconvex shell in which the high ventral interarea is divided medially by a ridge-like pseudodeltidium, a rectimarginate anterior commissure, and in the absence of colleplax which is present in Matutella , although in a rudimentary form (Andreeva 1962, pi. 5, fig. le). Acareorthis profunda sp. nov. Text-figures 2a-j, 3 Derivation of name. From the Latin profundus , deep. Holotype. Ventral valve, MGUH23740 (L 7-68, Lk 6-4. W 7-36, T 3-52) from the Paralleldal Formation (GGU collection 274907), Peary Land, central North Greenland. Paratypes. Figured: ventral valves, MGUH23741 (L 6 24, Lk 6 08. W 6 4, T 3 84); dorsal valves, MGUH23745 (L 5-84. W 6-8, Iw 5-36); MGUH23746 (L 4-56, W 5 04. Iw 40). Unfigured; MGUH23742 (L 6-72, Lk 608, W 6-88, T 3 04); MGUH23744 (L 4-56, W 4-88, Iw 3-36). Total of one complete shell, nine ventral, and three dorsal valves. All specimens from the same collection and locality as the holotype. 340 PALAEONTOLOGY, VOLUME 40 text-fig. 2. a-j, Acareorthis profunda sp. nov. a-d, holotype, MGUH23740; ventral valve in posterior, interior, exterior and lateral view respectively; x 5. e-f, MGUH23746; dorsal valve exterior, interior; x 5-2. g-h, MGUH23745; dorsal valve exterior, interior; x5-2; i-j, MGUH23741; ventral valve lateral view, posterior view; x 7. k-l, Pelmanella borealis gen. et sp. nov.; dorsal valve exterior, k, MGUH23750; l, MGUH23751 ; both x 5-6. All specimens from the Paralleldal Formation (late Early Cambrian), Peary Land, central North Greenland (GGU collection 274907). table 1. Acareorthis profunda sp. nov.; average dimensions of the ventral valves. L Lk W T L/W Lk/W T/L N 5 5 5 5 5 5 5 X 6-4 5-7 7-0 3-6 1 14% 81% 88% s 0-84 103 0-33 0-65 12 6 6-6 6-3 Min 5-6 4-5 5-8 3-4 95% 67% 46% Max 7-7 6-4 7-4 4-3 129% 100% 75% Diagnosis. Large for genus, ventral valve sub-pyramidal with maximum height at umbo; narrow sulcus originating near beak; umbonal perforation small, rounded; dorsal valve almost flat, lacking sulcus; radial ornament of 14-16 rounded costae. Description. Shell planoconvex; transversely rectangular in outline; posterior margin straight, occupying about 82 per cent, of maximum shell width. Ventral valve strongly convex, with shallow, but well-defined sulcus POPOV ET A L. : EARLY CAM BRIAN BRACHIOPODS 341 text-fig. 3. Acareorthis profunda sp. nov.; schematic drawings showing location of measurements, a, ventral valve exterior; b, ventral valve lateral view; c, ventral valve posterior; D, dorsal valve exterior. originating near beak ; sub-pyramidal, on average 1 14 per cent, as long as wide, and 55 per cent, as high as long, with maximum height at umbonal area; ventral beak pointed, with sub-circular perforation about 015-0-20 mm across, anterior to umbo; ventral interarea high, triangular, planar, apsacline, divided medianlly by narrow, ridge-like pseudodeltidium. Dorsal valve flat, plate-like, lacking sulcus. Shell coarsely costate with a maximum of 14-16 low, rounded costae. Ventral interior with thickened umbonal area. Dorsal interior with thickened ridge along posterior margin. Muscle scars and mantle canals of both valves not known. Remarks. A. profunda is comparable to A. jelli Roberts (in Roberts and Jell 1990) in having a planoconvex shell with a sub-pyramidal ventral valve. In both species the high, apsacline interarea is divided medially by a narrow, convex pseudodeltidium. However, it can be distinguished from the type species by its much larger size, the well-defined radial ornamentation with fewer costae, and its well-defined ventral sulcus. In addition, the ventral umbonal perforation is relatively small and a dorsal sulcus is absent. Roberts (in Roberts and Jell 1990, p. 270) noted the presence of a low dorsal interarea in A. jelli but our observations on topotypes of this species (kindly supplied by J. Roberts), as well as comparisons with our specimens from Greenland, suggest that the dorsal valve of Acareorthis is characterized by hemipheral growth and seemingly lacks an interarea. Order naukatida Popov and Tikhonov, 1990 Diagnosis. Shell biconvex, smooth or with radial ornament; ventral interarea with concave pseudodeltidium, which may be perforated by an elongate, sub-oval foramen; ventral visceral platform high, may be free peripherally; articulation with pair of closely spaced, ventral denticles in some genera, located on arcuate plate (anterise), and dorsal sockets on lateral sides of notothyrial platform; dorsal adductor scars arranged radially. Remarks. Naukatida was established by Popov and Tikhonov (1990) as a separate order, based mainly on the presence of a highly raised ventral muscle platform, and the unusual articulation present in Oina and Naukat. The articulation of both these genera consists of an anterise (that is, an arcuate plate anterior to the delthyrial margins) bearing paired denticles and a pair of sockets on the lateral sides of a highly raised notothyrial platform (Text-fig. 5). Pelmanella gen. nov. and Bynguanoia Roberts (in Roberts and Jell 1990) also exhibit these distinctive naukatid features, but differ in details of articulation and in the arrangement of muscle scars; they are referred here to the new family Pelmanellidae, described below. 342 PALAEONTOLOGY, VOLUME 40 text-fig. 4. a-h, Kutorginci ciiigulata (Billings, 1861). a, d-e, ventral valve pseudointerarea; a, MGUH23763; d, MGUH23753; E, MGUH23762; x 5. b, MGUH23760; juvenile ventral valve; x5; c, f, MGUH23759; dorsal valve interior, posterior view showing dorsal hinge ridges and furrows; x4; g, MGUH23754, dorsal valve interior; x 4. h, MGUH23761 ; juvenile ventral valve; x 5. i-l, Kutorginci catenata Koneva, 1979; i, PMKg3; ventral valve interior showing ventral hinge grooves on the lateral sides of pseudodeltidium; x 10. j, CNIGR 23/12589; ventral valve posterior view; x 3. k-l, CNIGR 22/12589; ventral valve exterior, lateral view; x3. m-p, Pelmanella borealis gen. et sp. nov.; m-n, dorsal valve interior; m, MGUH23749; n, MGUH23747; both x 4. o-p, ventral valve interior; o, MGUH23743, holotype; p, MGUH23748; both x 4. POPOV ET AL. \ EARLY CAMBRIAN BRACHIOPODS 343 Superfamily naukatoidea Popov and Tikhonov, 1990 Family pelmanellidae fam. nov. Diagnosis. Shell with rudimentary dorsal interarea, lacking notothyrial platform; ventral interior lacking denticles on anterise; posterior adductor scars on separate paired cardinal muscle platforms in both valves. Genera assigned. Pelmanella, Lower Cambrian (Toyonian), North Greenland; Bynguanoia , lower Middle Cambrian (Ordian), New South Wales, Australia. Remarks. Pelmanella and Bynguanoia both possess an anterise and high ventral muscle platforms and can be referred to the naukatides. The genera differ, however, from Oina and Naukat (and the family Naukatidae) in having a rudimentary, undivided dorsal interarea, paired cardinal muscle platforms in both valves, and in lacking paired denticles on the anterise. It is likely that the anterise in Pelmanella and Bynguanoia may have served as an articulatory structure in fixing the inner margins of the raised dorsal cardinal platform. Dorsal muscle scars posterior to the rotational axis that may have served as diductors are unknown. Conceivably, the shell opening mechanism was hydraulic, and may have been activated by outside lateral muscles (on the outer parts of the ventral cardinal muscle platform; Text-fig. 5) attached to the anterior body wall. In contrast, members of the Naukatidae have paired denticles on the anterise and possessing a notothyrial platform with rudimentary, paired sockets, and may have had oblique muscles serving as diductors (Popov and Tikhonov 1990). Bojarinovia Aksarina (in Aksarina and Pelman, 1978) and Swantonia Walcott (see Rowell 1977) may also belong within the Pelmanellidae, but their internal morphologies are unknown. Genus pelmanella gen. nov. Derivation of name. After the late Dr Ju. L. Pelman, a pioneer in the study of Early Cambrian brachiopods who died prematurely in a field accident. Type species. Pelmanella borealis sp. nov. Diagnosis. Shell slightly ventribiconvex, elongate, sub-oval to sub-circular in outline, smooth; anterior commissure rectimarginate; ventral valve with rudimentary apsacline interarea; delthyrium open, narrow, triangular, with distal margins joined by anterise; dorsal valve gently convex with rudimentary interarea; ventral interior with paired cardinal muscle platforms; ventral muscle platform separated from cardinal platforms by deep, oblique grooves; dorsal interior with paired cardinal muscle platforms, bearing medially located anterior adductor scars, divided by low median ridge. Remarks. This genus resembles Oina (Popov and Tikhonov 1990) in having a smooth, slightly ventribiconvex, sub-circular shell. Pelmanella is distinguished from the latter in having an open delthyrium, paired cardinal muscle platforms in both valves, and a weak dorsal median ridge. In addition, a notothyrial platform and denticles are lacking. In the internal morphology of both valves Pelmanella is closely similar to Bynguanoia Roberts (in Roberts and Jell, 1990), but it differs a— h, m-p from the Paralleldal Formation (late Early Cambrian), Peary Land, central North Greenland (GGU collection 274907); i-l from the Lower Cambrian (Botomian), south Tien-Shan, Alaj Range, Chachme River (locality 5069-8 of Popov and Tikhonov 1990). 344 PALAEONTOLOGY, VOLUME 40 table 2. Pelmanella borealis gen. et sp. nov.; average dimensions of dorsal valves. L W Cl Cw SI L/W Sl/L N 8 8 4 4 5 8 5 X 4-9 5-2 1-6 3-3 3-2 94% 51% s 1-71 1-61 056 102 1-38 5-4 13-5 Min 2-7 3-0 0-5 2-3 1-76 87% 41% Max 8-0 8-0 2-2 4-6 4-46 100% 71% posterior adductor text-fig. 5. Pelmanella borealis gen. et sp. nov.; schematic drawings showing location of measurements and position of muscle scars. A, ventral valve interior; b, dorsal valve interior. table 3. Kutorgina cingulata (Billings, 1861); average dimensions of ventral valves. L W T Iw Pw L/W T/L Iw/W Pw/Iw N 20 20 20 19 17 20 20 19 17 X 8-68 9-82 2-89 7-68 5-74 88% 33 % 80% 76% s 1-847 1-981 0-871 1-632 1-247 4-9 8-5 6-8 8-2 Min 4-5 4-6 1-5 4-2 2-9 76% 24% 67% 62% Max 12-4 130 4-8 1 15 8-5 97% 65% 90% 93% POPOV ET AL.\ EARLY CAMBRIAN BRACHIOPODS 345 table 4. Kutorgina cingulata (Billings, 1861); average dimensions of dorsal valves. L W Iw Pw L/W Iw/W Nw/Iw N 7 7 7 6 7 7 6 X 7-02 9-34 6-94 5-48 75% 75% 79% s 1-628 1-760 1470 1048 7-2 12-7 46 Min 4-5 6-5 4-6 3-8 64% 59% 74% Max 8-9 1 14 8-8 6-6 86% 95% 86% from the latter in having a smooth shell with a rectimarginate anterior commissure, strongly reduced ventral propareas, and in the absence of a pseudodeltidium. Pelmanella borealis sp. nov. Text-figures 2k-l, 4m-p, 5 Derivation of name. From the Latin borealis , northern. Holotype. MGUH23743, ventral valve (L 7-27, W 712, Cl 2-56, Cw 3-36, MP1 2-88, MPw 248, Iw 3 04) from the Paralleldal Formation (GGU collection 274907), Peary Land, central North Greenland. Paratypes. Figured : ventral valve, MGUH23748 (L 712, W 6 8, Cl 2 4, Cw 3 44, MP1 3 36, MPw 2-24, Iw 2 08); dorsal valves, MGUH23747 (L 8-0, W 8 0, Cl 2-24, Cw 4-64, SI 4-96); MGUH23751 (L 544, W 5 52, Cl L92, Cw 2-72, SI 3-36) ; MGUH23749; MGUH23750. Total of three ventral and 14 dorsal valves. All specimens from the same collection and locality as the holotype. Diagnosis. As for genus. Description. Ventral valve moderately and evenly convex, about 102-105 per cent, as long as wide, with somewhat acuminated beak; ventral interarea rudimentary, apsacline, occupied mainly by narrow, triangular delthyrium, with distal margins joined by anterise. Dorsal valve gently convex, sub-circular, about 94 per cent, as long as wide; dorsal interarea rudimentary, divided medially by narrow, shallow groove. Ventral interior with strongly raised, solid, antero-median muscle platform, occupying about 40-48 per cent, of maximum valve length ; ventral cardinal muscle platform with scars of the posterior adductors and, possibly, oblique lateral muscles; ventral cardinal muscle field separated from anterior muscle platform by deep oblique groove. Dorsal interior with posterior adductors situated on elevated posterolateral muscle platforms; dorsal anterior adductor scars large, elongate sub-oval, weakly impressed; dorsal median ridge low, extending anteriorly to mid-valve. Order kutorginida Kuhn, 1949 Diagnosis. Shell inequibiconvex to planoconvex, anterior margin rectimarginate, rarely sulcate; posterior margin wide, straight, with large median opening; delthyrium widely triangular, covered by convex pseudodeltidium, bounded laterally by furrows; beak with small, rounded apical foramen; dorsal interarea divided by wide notothyrium; both valves with slightly thickened, weakly defined visceral area, situated close to posterior margin; dorsal diductor scars placed on floor of notothyrial cavity; cardinal process absent; articulation without teeth and dental sockets; digestive tract probably open with anus placed posteromedianly. 346 PALAEONTOLOGY, VOLUME 40 Superfamily kutorginoidea Schuchert, 1893 Diagnosis. Shell with articulation characterized by two triangular plates formed by dorsal propareas and bearing oblique ridges on the inner sides, which seat into deep furrows formed by ridges along inner sides of ventral propareas and lateral extensions of pseudodeltidium. Family kutorginidae Schuchert, 1893 (inch Yorkiidae Rowell, 1962 and Agyrekiidae Koneva, 1979) Diagnosis. As for superfamily. Remarks. Recent studies of the articulation of Kutorgina and Nisusia (Rowell and Caruso 1985; Popov and Tikhonov 1990) suggest that kutorginids may represent one of the most primitive types of articulate brachiopod. Their hinge mechanism is simple and quite different from that present in most other Palaeozoic brachiopods; it consists of two narrowly triangular propareas in the dorsal valve, fitting into two deep furrows bounding the margins of the pseudodeltidium in the ventral valve. This type of articulation is also present in the Agyrekiidae and Yorkiidae, and there seems to be little reason to keep these families separated from the Kutorginidae. Genus kutorgina Billings, 1861 Type species. Kutorgina cingulata Billings, 1861, p. 8; Lower Cambrian, L’Anse au Loup (Belleisle), Labrador. Diagnosis. Shell strongly ventribiconvex to planoconvex; strongly lamellose peripherally, with granular micro-ornament; ventral interarea apsacline to orthocline, occupied mainly by broad, convex pseudodeltidium, bordered by deep hinge grooves; dorsal propareas narrowly triangular with deep dorsal hinge grooves and strong hinge ridges on the inner sides; dorsal interior with radially arranged adductor scars and diductor scars situated on low, transversely sub-triangular notothyrial platform; dorsal mantle canals pinnate. Remarks. The genus is in need of revision, but this is outside the scope of this paper, even though we assign specimens to the type species. A major problem is that species of Kutorgina have very few distinctive characters, and details of ornamentation, outline, and shell profile, as well as other EXPLANATION OF PLATE 1 Figs 1-12. Kutorgina cingulata (Billings, 1861). 1, MGUH23753; ventral pseudointerarea; x 5. 2-3, MGUH23752; ventral valve, posterior view, exterior; x 4. 4-5, MGUH23757, dorsal valve; 4, posterior view, x 5; 5, exterior, x 3. 6, MGUH23755; dorsal pseudointerarea showing dorsal hinge ridges and hinge furrows; x 5. 7-9, MGUH23756; ventral valve; 7, posterior view, 8, exterior, 9, lateral view; all x 3. 10-12, MGUH23758; ventral valve; 10, lateral view, 11, ventral interarea, 12, exterior; all x 3. All from the Paralleldal Formation (late Early Cambrian), Peary Land, central North Greenland (GGU collection 274907). Figs 13-18. Kutorgina catenata Koneva, 1979. 13-15, PMKgl ; ventral valve, 13, exterior, 14, lateral view, 15, oblique posterior view; all x 1 1. 16, PMKg2; dorsal valve interior; x 10. 17-18, PMKg4; juvenile dorsal valve; 17, exterior, x9; 18, posterior view, x 10. All from the Lower Cambrian (Botomian), south Tien- Shan, Alaj Range, Chachme River (locality 5069-8 of Popov and Tikhonov 1990). PLATE 1 POPOV et al., Kutorgina 348 PALAEONTOLOGY, VOLUME 40 morphological features seem to vary strongly throughout ontogeny and within populations. As a result, the majority of the species described is not well-defined. Kutorgina cingulata (Billings, 1861) Plate 1, figures 1-12; Text-figure 4a-h 1912 Kutorgina cingulata (Billings) Walcott, p. 580, pi. 5, figs 1, I a-e [see for synonymy], Lectotype. Selected here from the syntypes (deposited in the Geological Survey of Canada, Ottawa) described by Billings (1861), GSC 384a, ventral valve (L 15-8, W 17-3, T 5-0); from the Lower Cambrian of L’Anse au Loup, Labrador. Material. Figured, from the Paralleldal Formation (GGU collection 274907), Peary Land, central North Greenland; MGUH23752-23763. Total of 74 ventral and 29 dorsal valves. Diagnosis. Shell transversely sub-rectangular in outline; ventral interarea strongly apsachne to orthocline in adults; ventral sulcus narrow, originating near umbo, shallowing anteriorly; dorsal valve flat with convex umbonal area; notothyrium broadly triangular with vestigial apical chilidium; ventral interior characters poorly known. Description. Shell ventribiconvex to planoconvex, transversely sub-rectangular in outline, on average 89 per cent, as long as wide, with maximum width at mid-length. Ventral valve moderately to strongly and evenly convex, on average 33 per cent, as high as long; shallow sulcus originating at umbonal area, usually disappearing anterior to mid-valve; ventral interarea apsachne in juveniles, orthocline in adults, with broadly triangular, convex pseudodeltidium and rudimentary propareas, separated from pseudodeltidium by deep, widely divergent grooves; umbo pointed, bearing small apical foramen. Dorsal valve flattened with raised umbonal area; notothyrium forming broad triangular notch, bordered laterally by grooved triangular propareas and covered apically by rudimentary chilidium (Text-fig. 7a-c); dorsal hinge furrows on inner sides of propareas bordered by hinge ridges which, in some large specimens, are comparable to the socket plates of Nisusia. Ornamented with numerous, slightly irregular, imbricating lamellae, strongly developed in ventral valve, but usually weakly defined in dorsal valve. Muscle scars and mantle canals weakly impressed in both valves. Dorsal umbonal area with short, slightly raised notothyrial platform; dorsal anterior and posterior adductor scars radially arranged. Mantle canals poorly defined; proximal parts of pinnate mantle canals observed in one dorsal valve. Remarks. The Greenland specimens are indistinguishable from the types of K. cingulata in most morphological characters. However, they are generally smaller than those from North America (maximum width 30 mm; Walcott 1912, p. 580) and lack the distinctive granular micro- ornamentation, but this is probably due to the secondary silicification of the shell. K. reticulata Poulsen, 1932, from the Early Cambrian Ella Island Formation of north-east Greenland, seems to differ from K. cingulata mainly in the absence of a dorsal median fold. K. cingulata is distinguished from K. catenate! Koneva, 1979, from the Lower Cambrian of Kazakhstan and Kirgizia, in having a ventral sulcus, a well-defined notothyrial platform, and strongly developed ridges bounding the inner sides of the grooves in the dorsal pseudointerarea. Trematosia undulata (Cooper, 1976), from the Early Cambrian Nimra Formation of southern Negev, Israel, is also similar to K. cingulata in its strongly lamellose shell, minute apical foramen, broadly triangular, convex pseudodeltidium, and poorly defined propareas on an apsachne ventral interarea. T. undulata is here re-assigned to Kutorgina , but it differs from K. cingulata in having a gently convex ventral valve lacking a sulcus. The wide geographical distribution of Kutorgina in the late Early Cambrian is comparable to the cosmopolitan distribution of acrotretoids (e.g. Hadrotreta , Linnarssonia ) and acrothelids (e.g. Eothele , Acrothele ) and may indicate that the larvae of Kutorgina (and possibly other kutorginides) were planktotrophic, unlike most other calcareous brachiopods. POPOV ET AL.\ EARLY CAMBRIAN BRACHIOPODS 349 Kutorgina catenate i Koneva, 1979 Plate 1, figures 13-18; Text-figure 4i-l 1979 Kutorgina catenata Koneva, p. 58, pi. 23, figs 2-8; pi. 24, figs 1-5. 1990 Kutorgina catenata Koneva; Popov and Tikhonov, p. 45, pi. 3, figs 13-16; pi. 4, figs 7-9. Holotype. Ventral valve (2138/226, Institute of Geological Sciences, Alma-Ata); from an Early Cambrian limestone olistolith, Agyrek Mountains, Kazakhstan. Remarks. Silicified specimens of Kutorgina catenata from the Lower Cambrian of south Tien-Shan, Kirgizia are illustrated here for comparison with K. cingulata. Genus agyrekia Koneva, 1979 Type species. Agyrekia alata Koneva, 1979; Lower Cambrian, Agyrek Mountains Kazakhstan, by original designation. Diagnosis. Shell sub-equibiconvex, ornamented by concentric rugellae; ventral interarea catacline to procline, with well-defined narrow, flattened propareas; pseudodeltidium broad, gently convex; poorly defined ventral hinge grooves on lateral sides of pseudodeltidium; dorsal valve moderately convex; ventral mantle canals pinnate. Remarks. In the original diagnosis of Agyrekia (Koneva 1979, p. 60) the dorsal valve was described as being high, sub-conical, with a well-developed homeochilidium. However, nearly all illustrated ‘dorsal’ valves of the type species A. alata (Koneva 1979, pi. 26, fig. 4; pi. 27, figs 2, 4; pi. 28, fig. 2; pi. 29, figs 2-3) are indistinguishable from the ventral valves in the morphology of the interarea, and clearly show shallow, but distinctive, hinge grooves on the lateral sides of the ‘homeochilidium’, thus indicating that they are ventral valves. Consequently, the dorsal valve of A. alata is known only from a single incompletely preserved specimen (Koneva 1979, pi. 26, fig. 1). Agyrekia differs from Kutorgina in having a much more strongly convex dorsal valve, with the maximum height anterior to the umbo. The morphology of the dorsal interarea is unknown in the type species, but it is likely that the dorsal valve of Agyrekia had a broad, sub-triangular, open notothyrium and short, triangular grooved propareas like other kutorginids. In particular, this type of morphology can be observed in the dorsal valves from Greenland which are assigned here to Agyrekia all. obtusa Koneva. Agyrekia is closely similar to Kutorgina in the morphology of the ventral and dorsal interareas which have primitive articulatory structures. In Agyrekia aft', obtusa , however, the ventral hinge grooves on the lateral sides of the pseudodeltidium are weakly developed, the axis of rotation is fixed by the outer margins of the pseudodeltidium and the ventral propareas fit into furrows on the inner sides of the dorsal propareas. This pattern of the articulation is somewhat comparable to that of nisusiides, and the dorsal hinge ridges and furrows of Agyrekia aff obtusa may be homologous to the socket plates and sockets of Nisusia. Agyrekia aff. obtusa Koneva, 1979 Text-figure 6 Material. Figured: from the Paralleldal Formation (GGU collection 274907), Peary Land, central North Greenland; MGUH23766, ventral valve (L 1T5, W 14-6, T 4-2, Iw 10-4), MGUH23767; dorsal valves: MGUH23768 (L 7-4, W 9 0, T 2-3, Iw 6 4); MGUH23764 (L 7-7, W 10 0, T 2-96, Iw 6-6); MGUH23765 (L 7-4, W 9-2, Iw 5-2). Total of three ventral and five dorsal valves. Description. Shell slightly ventribiconvex; transversely sub-oval in outline, with maximum width at mid-length. Ventral valve sub-conical, with posterior and lateral slopes gently convex in cross section; about 79 per cent. 350 PALAEONTOLOGY, VOLUME 40 text-fig. 6. Agyrekia aff. obtusa Koneva, 1979. a-c, MGUH23766; ventral valve exterior, lateral view, posterior view; all x2-8. d-e, MGUH23767; ventral valve. D, exterior; x2-8; e, pseudointerarea; x 4-2. f-g, MGUH23764; dorsal valve exterior, lateral view; x 4-2. h-i, MGUH237688; dorsal valve lateral view, exterior; both x 5-6. J, MGUH23765; dorsal valve interior; x4-2. All specimens from the Paralleldal Formation (late Early Cambrian), Peary Land, central North Greenland (GGU collection 274907). as long as wide, and 37 per cent, as high as long, with maximum height at apex; perforated by small, rounded foramen; ventral interarea high, planar, procline, with broad, sub-triangular, gently convex pseudodeltidium, bordered by widely divergent grooves; ventral propareas narrow, flattened, well-defined. Dorsal valve moderately and evenly convex, about 80 per cent, as long as wide; ventral interarea rudimentary anacline, with narrow hinge ridges and furrows on inner sides of propareas; notothyrium broadly triangular, occupying about 81 per cent, of width of posterior margin. Shell smooth with fine, slightly irregular growth lamellae; micro-ornamentation not preserved. Muscle scars and mantle canals weakly impressed in both valves. Remarks. Our specimens are closely similar to A. obtusa Koneva (1979, p. 62) with respect to the obtusely sub-conical ventral valve, the procline ventral interarea with gently convex pseudodeltidium, bordered by widely divergent grooves, and the narrow, flattened propareas. As noted above, all but one of the dorsal valves of A. obtusa illustrated by Koneva (1979, pi. 31, figs 3-4) are probably ventral valves. The type specimens of Kutorgina catenata , which come from the same locality as A. obtusa , include dorsal valves representing two different morphologies. In particular, one specimen POPOV ET A L.: EARLY CAMBRIAN BRACHIOPODS 351 illustrated by Koneva (1979, pi. 24, fig. 5) represents a typical Kutorgina , having a flattened dorsal valve, with the maximum height at the pointed, slightly elevated beak, whereas another specimen (pi. 24, fig. 3) has a moderately convex dorsal valve with the maximum height near mid-valve. The latter specimen is closely similar to the dorsal valves of A. aff. obtusa in our collection and most probably represents a dorsal valve of A. obtusa. Another similar species is Kutorgina pyramidalis Aksarina (see Aksarina and Pelman 1978, p. 86, pi. 8, figs 1-9) from the Lower Cambrian (Toyonian), Batenev Ridge, south-western Siberia. It has a subequally biconvex shell with a moderately convex dorsal valve and a procline ventral interarea. A. aff. obtusa differs from the latter species, however, in having a weakly developed lamellose ornamentation. K. pyramidalis differs from other species of Kutorgina in having a convex dorsal valve with the maximum height slightly posterior to mid-length, as well as a sub-pyramidal ventral valve with a procline interarea and well-defined propareas. It is provisionally re-assigned here to Agyrekia. A. aff. obtusa is also comparable to Kutorgina1. sp. described by Roberts and Jell (1990, p. 285) from the early Mid Cambrian (Ordian) First Discovery Limestone of New South Wales, Australia. The Greenland species differs in having poorly developed concentric lamellae and in lacking a ventral sulcus. REMARKS ON ARTICULATION IN EARLY CALCAREOUS BRACHIOPODS Rowell and Caruso (1985) demonstrated significant differences between the hinge mechanisms of nisusiides and the earliest orthides. Nisusia lacks teeth, and the sockets and sockets ridges on the lateral sides of the notothyrium are partly composed of primary shell. These are probably are not homologous to the sockets and brachiophores of primitive orthides. Popov and Tikhonov (1990) described primitive articulatory structures in Early Cambrian Kutorgina and Nisusia from south Kirgizia. They noted similarities in the articulation of these two genera and re-assigned Nisusia and the related genus Eoconcha to the order Kutorginida, forming the separate superfamily Nisusioidea. The articulation in Kutorgina cingulata consists of deep ventral hinge furrows which separate the pseudodeltidium from the narrow ventral propareas, as well as strong hinge ridges, bounding the hinge furrows on the inner parts of triangular dorsal propareas, along the sides of the notothyrial opening (Text-fig. 7). All these structures are also present in Kutorgina catenata (P!. 1, fig. 16; Text- fig. 4i), but the dorsal hinge ridges are strongly developed in K. cingulata and have some similarity to the socket plates of Nisusia (PI. 1, fig. 6; Text-fig. 4f-g). The axis of rotation between the valves in Kutorgina was fixed at two points by the distal parts of narrowly triangular dorsal propareas, which served as ‘teeth’, fitting into the deep ventral hinge furrows, essentially functioning as ‘sockets'. The dorsal hinge ridges and furrows served as supporting structures in restricting the lateral movements of the valves. In Nisusia (Rowell and Caruso 1985; Popov and Tikhonov 1990), the rotation axis coincides with the posterior commissure; it is fixed by the lateral margins of the pseudodeltidium which fit into the deep sockets on the inner margins of dorsal interareas (Text-fig. 7). The socket plates and sockets in Nisusia are probably homologous in Kutorgina to the dorsal hinge ridges and furrows respectively. These structures are situated on the inner margins of dorsal propareas and are composed partly of primary shell. The main difference between the two genera is that the dorsal hinge ridges and furrows in Kutorgina have supplementary articulatory functions, whereas the socket plates and sockets in Nisusia are the main articulatory structures in the dorsal valve. The articulation in Kutorgina seems to be more primitive and may be present in a rudimentary state in juvenile shells of Nisusia (Popov and Tikhonov 1990, pi. 3, figs 23-24). The pattern of articulation in kutorginides (including nisusioids) was probably not ancestral to that of the early orthides (superfamily Protorthoidea). The earliest known protorthoidean genera (e.g. Glyptoria and Israeleria) completely lack structures comparable to the socket plates and sockets of Nisusia. The primitive brachiophores and sockets are composed entirely of secondary shell, and may have evolved later in relatively more derived taxa such as Arctohedra. 352 PALAEONTOLOGY, VOLUME 40 Kutorgina Nisusia B foramen - pseudodeltidium chilidium dorsal hinge ridge socket plate chilidium text-fig. 7. Schematic illustration of the articulation in kutorginids. a-c, Kutorgina , ventral valve interior, dorsal valve interior, posterior view, d-f, Nisusia , ventral valve interior, dorsal valve interior, posterior view (see also Holmer and Popov 1996, p. 13). The presence of hinge groves on the lateral sides of the pseudodeltidium together with narrow, triangular propareas in the dorsal valve of Yorkia wanneri (see Rowell 1962), suggests an articulation pattern similar to that of Kutorgina. The specimens of Trematosia sp. 1 figured by Cooper (1976, pi. 1, figs 39^41), as well as some specimens of Trematosia radifer (Cooper 1976, pi. 1, figs 19-26), are probably not congeneric with the holotype of Trematosia radifer because they have a minute apical foramen, along with a kutorginide-like morphology; possibly they may be referred to Yorkia. The holotype and several paratypes of Trematosia radifer (Cooper 1976, pi. 3, figs 1-6, 32) differ significantly from the other specimens referred to Trematosia in that they have a large foramen anterior to the apex, along with a poorly defined ventral interarea and a different dorsal interarea, probably with some primitive hinge structures. These features are unknown in kutorginides, and the affinities of these taxa with other early calcareous brachiopods remain uncertain. Acknowledgements. We are grateful to John Laurie (Canberra, Australia) and John Roberts (Kensington, N.S.W., Australia) who supplied comparative material from Australia. This work has been supported by several grants (to L. Holmer and .1 S. Peel) from the Swedish Natural Science Research Council (NFR). POPOV ET AL.\ EARLY CAMBRIAN BRACHIOPODS 353 Leonid Popov gratefully acknowledges receipt of a one year NFR visiting scientist grant, as well as three visiting scientist grants from the Royal Swedish Academy of Sciences (KVA), which enabled him to work at the Institute of Earth Sciences, Department of Historical Geology and Palaeontology, Uppsala University. For the past decade Bert Rowell's research has been supported by the Office of Polar Programs, National Science Foundation, currently by grant OPP91- 17444. Publication is authorized by the Geological Survey of Greenland. Copenhagen. Denmark. REFERENCES aksarina, N. a. and pelman, ju. l. 1978. Kembrijskije brakhiopody i dvustvorchatye molljuski Sibiri. [Cambrian brachiopods and bivalved molluscs of Siberia.] Trudy Institute i Geologii i Geofiziki , 362, 5-178. [In Russian], andreeva, o. N. 1962. Nekotoryje kembrijskije brakhiopody Sibiri i Srednej Asii. [Some Cambrian brachiopods of Siberia and Central Asia.] Paleontologicheskij ZhurnaL 2 (1962), 87-96. [In Russian]. billings, e. 1861. On some new or little known species of Lower Silurian fossils from the Potsdam group (Primordial zone). Report on the Geology of Vermont , 2 (appendix), 942-960. COOPER, G. A. 1951. New brachiopods from the Lower Cambrian of Virginia. Journal of the Washington Academy of Sciences , 41, 4-8. — 1976. Lower Cambrian brachiopods from the Rift Valley (Israel and Jordan). Journal of Paleontology, 50, 269-289. debrenne, f. and peel, J. s. 1986. Archaeocyatha from the Lower Cambrian of Peary Land, central North Greenland. Rapport Gronlands Geologiske Undersogelse, 132, 39-50. higgins, a. K., ineson, j. R., peel, J. s., surlyk, f. and sonderholm , m. 1991. Lower Palaeozoic Franklinian Basin of North Greenland. Bulletin Gronlands Geologiske Undersogelse , 160. 71-139. holmer, l. e. and popov, l. e. 1996. De kalkskaliga brachiopodernas tidigaste evolution - nya fynd fran Kirgisistan. Geologiskt Forum, 9. 10-14. ineson, j. r. and peel, j. s. in press. Cambrian stratigraphy of North Greenland. Bulletin Gronlands Geologiske Undersogelse. koneva, s. p. 1979. Stenotekoidy i bezzamkovye brakhiopody nizhnego i nizov srednego kembrija Tsentralnogo Kazakhstana. [Stenothecoids and inarticulate brachioipods front the Lower and lower Middle Cambrian of central Kazakhstan.] Nauka, Alma-Ata, 123 pp. [In Russian], kuhn, o. 1949. Lehrbuch der Paldozoologie. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart, 326 pp. peel, j. s. and sonderholm, m. 1991. Sedimentary basins of North Greenland. Bulletin of the Geological Survey of Greenland, 160. 1-164. pelman, Ju. l 1977. Ranne- i srednekembriiskie bezzamkovye brakhiopody Sibirskoi platformy. [Early and Mid Cambrian inarticulate brachiopods from the Siberian Platform]. Akademiia Nauk SSSR, Sibirskoe Otdelenie, Institut Geologii i Geofiziki ( IGIG ), Trudy ( Novosibirsk ), 316, 1-168. [In Russian]. popov, l. e. and tikhonov, Yu. a. 1990. Rannekembriiskie brakhiopody iz yuzlmoi Kirgizii. [Early Cambrian brachiopods from southern Kirgizia]. Paleontologiclteskii Zhurnal, 3 (1990), 33 — 46 [In Russian]. poulsen, c. 1932. The Lower Cambrian faunas of East Greenland. Meddelelser om Gronland, 87, 1-67. Roberts, j. and jell, p. a. 1990. Early Middle Cambrian (Ordian) brachiopods of the Coonigan Formation, western New South Wales. Alcheringa, 14. 257-309. rowell, a. J. 1962. The genera of the brachiopod superfamilies Obolellacea and Siphonotretacea. Journal of Paleontology, 36, 136-152. — 1965. Class Inarticulata. H260-H296. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part H. Brachiopoda 1(2). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 927 pp. 1977. Early Cambrian brachiopods from the southwestern Great Basin of California and Nevada. Journal of Paleontology, 51, 68-85. — and caruso, n. e. 1985. The evolutionary significance of Nisusia sulcata, an early articulate brachiopod. Journal of Paleontology, 59, 1227-1242. schuchert, c. 1893. A classification of the Brachiopoda. The American Geologist, 11, 141-167. walcott, c. D. 1912. Cambrian Brachiopoda. Monograph of the US Geological Survey, 51, 1-872. LEONID E. POPOV VSEGEI St Petersburg 199 026, Russia 354 PALAEONTOLOGY, VOLUME 40 LARS E. HOLMER JOHN S. PEEL. Institute of Earth Sciences Department of Historical Geology and Palaeontology Norbyvagen 22, S-752 36 Uppsala, Sweden ALBERT J. ROWELL Department of Geology Typescript received 11 January 1996 The University of Kansas Revised typescript received 22 July 1996 Lawrence, Kansas 66044, USA LATE JURASSIC BRACHIOPODS FROM NORTH-EAST IRAN by m. h. adabi and d. v. ager Abstract. A brachiopod fauna from the Mozduran Formation of the Kopet-Dagh Basin in north-east Iran, near the border with Turkmenistan, is described. It is of Late Jurassic age and shows affinities with European forms, especially those of 'Boreal’ type from Russia. There are no 'Tethyan’ or ‘Ethiopian’ forms present. The Kopet-Dagh basin in north-east Iran is about 600 km long and about 200 km wide. It was established after mid Triassic orogenic movements, when the Iran and Turan plates had apparently joined (Berberian and King 1981). From mid Jurassic times onwards, the basin was invaded by a widespread Mesozoic epicontinental sea. It subsided and acquired a thick sequence (about 10 km) of almost continuous shallow marine to continental sediments ranging from Jurassic to Oligocene in age, with no major sedimentary breaks or volcanic activity. The basin started subsiding along major longitudinal faults during the Jurassic (Berberian and King 1981). Moussavi-Harami and Brenner (1992), however, suggested that much of the post-Jurassic subsidence in the eastern Kopet- Dagh basin was caused by sediment loading rather than tectonism. No important orogenic movements took place after early Jurassic times until the area was folded late in the Alpine Orogeny along south-east to north-west lines. The study area is located to the north-east of Mashhad (Text-fig. 1) between latitudes 30° and 36° 10' N and longitudes 60° 30' and 61° E. The general stratigraphy of the Kopet-Dagh basin in this area comprises 15 different formations (Text-fig. 2) from mid Jurassic to Oligocene age (Afshar- Harb 1970; Kalantari 1987). The brachiopods described herein are from the Upper Jurassic Mozduran Formation, which is exposed in the Sarakhs area of north-east Iran. It conformably overlies fluvio-deltaic to shallow marine deposits of the Kashafrud Formation (Madani 1977) and is succeeded conformably by the Shurijeh fluvial siliciclastics (Moussavi-Harami 1986). For- aminiferal studies by Kalantari (1969) indicate that the Mozduran formation ranges in age from Callovian to Kimmeridgian. The brachiopods were collected from the upper part of the Mozduran section. At the time of formation the basin was situated probably at a palaeolatitude of about 20° N (Smith et at. 1981). The Mozduran carbonates consists of diverse skeletal and non-skeletal grains, with abundant red and blue-green algae, evaporites and early diagenetic dolomites. Petrographic and geochemical evidence indicates that the Mozduran Formation was deposited in a warm, tropical, shallow marine environment (Adabi and Rao 1991). The purpose of the paper is to describe this new brachiopod fauna and to examine its stratigraphical and biogeographical implications. SYSTEMATIC PALAEONTOLOGY Order terebratulida Waagen, 1883 Superfamily terebratuloidea Gray, 1840 Family terebratulidae Gray, 1 840 Subfamily loboidothyrinae Makridin, 1964 Genus moeschia Boullier, 1976 (Palaeontology, Vol. 40, Part 2, 1997, pp. 355-362, 1 pi.) © The Palaeontological Association 356 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Location of sections in Kopet-Dagh Basin, north-eastern Iran. Remarks. This genus is common in the Upper Jurassic of Europe. Besides the type species, Terebratula alata Rollet, Boullier (1976) also included Loboidothyris zeiteni (de Loriol). The latter was well known to one of us (DVA) and occurs commonly in the Kimmeridgian of the French Jura (Ager and Evarny 1963). It is found right across Europe to Poland (Barczyk 1969) and the Russian platform (Makridin 1964) and is closely related to the form described here. Buckman (1918) originally described Loboidothyris from the Aalenian and Bajocian but Makridin (1960) recorded it in the Oxfordian of Russia, and Ksiazkiewicz (1974) recorded it doubtfully in Poland from as high as the Tithonian. Boullier ( 1976) described Moeschia species from the Upper Jurassic of France, and the Iranian form described below can be attributed to M. subsella. Moeschia subsella (Leymerie, 1846) Plate 1. figures 1^1 Description. This is a small to medium-sized Moeschia , longitudinally oval in outline but with the characteristic truncated anterior margin. The largest specimen from Iran (albeit slightly damaged) is 41 mm long, 37 mm wide and 19 mm thick. The anterior commissure is uniplicate to gently biplicate, generally the latter. The beak is massive, erect and mesothyrid; the beak-ridges are rounded. The valves are biconvex of plano-convex. They are smooth with no discernible growth-lines. Makridin (1964) recorded this species from the Upper Oxfordian of the Russian platform, it is certainly close to M. zieteni , which ranges up through the Kimmeridgian to near the top of the Jurassic. There are 13 specimens in the Iranian collection which can be referred to M. subsella. Genus uralella Makridin, 1960 Remarks. This is one of a group of very large terebratulids that characterize the uppermost Jurassic. Besides Uralella , it includes Gigantothyris , Boreiothyris , Taimyrothyris , Siberiothyris, Juralina and Rouillieria. Although particularly characteristic of the Russian Platform (extending into Poland), Rouillieria is also found as far west as Britain (Ager 1971) as a member of a typically Boreal assemblage. The accompanying rhynchonellids, such as Torquirhynchia and Russirhynchia , are also often of an unusually large size. All these forms are characteristic of the Boreal region. ADABI AND AGER: JURASSIC BRACHIOPODS FROM IRAN 357 WEST EAST Uralella gigantea Makridin, 1964 Plate 1, figure 5 Description. All members of this genus are large but, as the name implies, this species is even larger than the rest. The single specimen from Iran (PI. 1, fig. 2) is the internal case of the pedicle valve and measures 67 mm long, 67 mm wide (though it is slightly broken and was probably a few millimetres wider) and about 21 mm deep for the single valve. From Makridin’s plates (1964, pis 18-19) U. gigantea appears to be equally biconvex, so the original depth of this specimen may have been about 42 mm. Makridin’s specimens are up to 84 mm long, 80 5 mm wide and 35 mm deep. His smallest specimen (Makridin 1964, p. 257) is 35 mm long, 37 mm wide and 1 9- 5 mm deep, but larger specimens appear to be the most common, and the small ones may be juveniles. The surface of the Iranian specimen is smooth with faint growth-lines, like those from Russia. The ventral beak is broken, but appears to be strongly incurved. As it is an internal cast, no punctae are visible. Makridin ( 1964) figured serial sections and a reconstruction, which show a short smooth loop turned back in a posterior direction, with a short connecting band. The teeth are long and slender, inserted obliquely in the sockets. No cardinal process is shown, but in other species of Uralella this is known to be thick with rather few ridges. 358 PALAEONTOLOGY, VOLUME 40 Remarks. Makridin recorded this species from the Lower Volgian. It is not very common in Russia, and Makridin (1964, p. 256) recorded only five well-preserved shells and four internal casts. Dagys (1968, p. 100) recorded 17 small damaged shells from the northern Urals. From the name it may be presumed that the genus comes from the Urals, though Makridin does not seem to be specific on this point. Dagys, however, mentions Yatriya in the northern Urals and again gives the horizon as Lower Volgian. Order rhynchonellida Kuhn, 1949 Superfamily rhynchonelloidea Gray, 1848 Family rhynchonellidae Gray, 1848 Subfamily cyclothyridinae Makridin, 1955 Genus torquirhynchia Childs, 1969 Remarks. Large late Jurassic rhynchonellids, commonly attributed in the past to Rhactorhynchia or to Septaliphoria. Their most obvious feature is their strongly asymmetrical anterior commissure, from which the name is derived. Torquirhynchia inconstans (J. Sowerby, 1816) Plate 1, figures 6-8 Description. This is the best known member of the genus and is particularly common in the Kimmeridgian at its type locality in Dorset, southern England. It is a large globose form with marked asymmetry in the anterior commissure. The largest specimens range up to about 35 mm long, 34 mm wide and 32 mm deep. The Iranian specimens do not reach such dimensions, but are comparable in every way to the type material. The beak is massive, erect and submesothyrid with fairly sharp beak ridges. There are up to 20 simple, unbranched costae of the tetrahedra type (Ager 1956) on each valve. They are usually equally divided between the two lateral halves of the shell. The shells are dextrally or sinistrally skewed in almost exactly equal numbers. It has been suggested that this represents sexual dimorphism, but there is not real evidence of this (Ager 1969). It seems more likely that it is an adaptation to functioning either the right way up in rough water, or whilst resting on one of the broad planareas. In the collection from Iran, 25 specimens appear to be attributable to this species, but it probably passes by insensible gradations into the following. Torquirhynchia lehmanni (Makridin, 1964) Plate 1, figures 9-1 1 Description. This is the most common brachiopod from the Mozduran Formation and 39 specimens belong here. The only apparent distinction from Torquirhynchia inconstans is the more depressed nature of the shells and the less marked nature of the asymmetry of the anterior commissure. An average specimen measures about 28 mm long, 32 mm wide and 17 mm deep, with the length to depth ratio much higher than in the type species. explanation of plate 1 Figs 1-4. Moeschia subsella (Leymerie, 1846); UMGD 9601 ; dorsal, lateral, anterior and ventral views; x 1-5. Fig. 5. Urale/la gigantea Makridin, 1964; UMGD 9602; ventral view of internal cast of pedicle valve; x 10. Figs 6-8. Torquirhynchia inconstans (J. Sowerby, 1816); UMGD 9603; dorsal, lateral and anterior views; x 1-5. Figs 9-1 1 . Torquirhynchia lehmanni (Makridin, 1964); UMGD 9604; dorsal, lateral and anterior views; x 1-5. Figs 12-14. Torquirhynchia speciosa (Munster, 1839); UMGD 9605; dorsal, lateral and anterior views, x 10. All anterior views have brachial valve uppermost. All specimens are from the Upper Jurassic, Mozduran Formation, upper part of Mozduran section, Sarakhs area, north-eastern Iran, and are housed in the Department of Geology, University of Mashhad, Iran. PLATE 1 ADABI and AGER, Moeschia, Uralella, Torquirhynchia 360 PALAEONTOLOGY, VOLUME 40 T. lehmanni was attributed by Makridin (1964, p. 109) to the closely related Upper Jurassic rhynchonellid genus Septaliphoria. Indeed it has been suggested (Ager et al. 1972, fig. 7) that Tor quirky nchia is a direct early offshoot of Septaliphoria. Makridin (1964) figured the species from the Lower Volgian of the Russian Platform. As the species is abundant in the Iranian collection, serial sections were made of one of the specimens to study its internal structures. These are illustrated in Text-figure 3. Points to note are the massive dental lamellae text-fig. 3. Serial transverse sections of Torquirhynchia lelmiawti (Makridin, 1964). The figures indicate the distance in millimetres from the posterior end of the shell. See text for discussion. in sections 2-5 to 3-5, the clear if short septalium supported by a dorsal median septum in section 4 0, the massive teeth inserted obliquely into the sockets in sections 5-5 to 6 0 and the slightly concave crura seen in sections 7-0, 7-5 and 8-0. These last strengthen the case for including the genus in the Cyclothyridinae. The only unusual features are the very short dorsal median septum and the thinness of the shell. The massive teeth and dental lamellae suggest adaptation to a shallow water, high energy environment, but this hardly fits with the thin shell. Perhaps one should not place too much confidence on such theorizing. It might be mentioned, however, that other large late Jurassic brachiopods were certainly adapted to living in high energy environments. Thus the huge terebratulid Juralina immanis (Zeuschner) lived in the reef environment of the Tithonian-Volgian Stramberk Limestone of the Czech Republic and was accompanied by the remarkable ‘ Rhynchonella ’ pachy theca Zeuschner, which is almost solid shell. Also in that assemblage is Loboido- thyrisl insignis (Schiibler) and an asymmetrical Septaliphoria asteriana (d’Orbigny). These were discussed in Ager (1965, pp. 153-156). A similar fauna was described from Wozniki in neighbouring Poland by Ksiqzkiewicz (1974). Torquirhynchia speciosa (Munster, 1839) Plate 1, figures 12-14 Description. Two specimens from Iran may be attributed to this distinctive species, which is characterized by its extremely wide shells. The larger of the two measures 38 mm long, 41 mm wide and is 23 mm deep. The ADABI AND AGER: JURASSIC BRACHIOPODS FROM IRAN 361 second specimen is smaller but of similar proportions. Strangely enough, one specimen is dorsally skewed and the other sinistrally. There are about 25 costae of the 'tetrahedra' type on each valve. Otherwise the external characters are similar to the other species, with a massive erect beak and fairly sharp beak ridges. Childs ( 1969) recorded and figured T. speciosa from the Upper Kimmeridgian to Lower Volgian near Ingolstadt in Germany. CONCLUSIONS The brachiopod assemblage from the Mozduran Formation clearly indicates a latest Jurassic age. Though brachiopods are not the best of fossils for stratigraphical purposes, the balance of evidence suggests an early Tithonian age, younger than that given by contemporary foraminifers. They also suggest a shallow water, high energy environment, and are particularly interesting from the palaeogeographical point of view. Their affinities are all with extra-Alpine Europe, and with the Russian Platform and Urals in particular. One notes, however, the absence of regular members of the comparable European assemblages of this age, such as Rhynchonella sensu stricto and Septaliphoria asteriana (d’Orbigny). Nevertheless, the fauna is very much one of the Boreal Province. This also contrasts with the suggestion given earlier in this paper of a tropical habitat, but this may be simply a matter of geographical connections rather than of actual latitude. Certainly there is no sign of Tethyan forms such as the pygopids and nothing from the Ethiopian Province, which extended as far north as Sinai and the southern part of the Arabian Peninsula at this time. It is concluded therefore that in late Jurassic times, Iran was part of the European Plate, as has been previously suggested for Turkey (Ager 1988) on the basis of similar brachiopod evidence. Acknowledgements. MHA acknowledges support provided by the Iranian Government and the Department of Geology, University of Tasmania, Australia. We thank Dr C. F. Burrett of the University of Tasmania, who acted as an intermediary in the study of this interesting fauna. Mr Frank Cross of the University College of Swansea made the excellent cellulose peels from which were drawn the sections shown in Text-figure 3. Both authors are indebted to Dr M. G. Bassett of the National Museum of Wales for arranging the specimens to be photographed by Mrs K. Bryant of that institution. We thank Dr P. Doyle for critically reading the earlier version of the manuscript and for making helpful suggestions. REFERENCES adabi, m. h. and rao, c. p. 1991. Petrographic and geochemical evidence for original aragonitic mineralogy of Upper Jurassic carbonates (Mozduran Formation), Sarakhs area, Iran. Sedimentary Geology , 72, 253-267. afshar-harjb, a. 1969. [A brief history of geological exploration and geology of the Sarakhs area and the Khangiran gas field]. Bulletin of the Iranian Petroleum Institute , 37, 86-96. [In Persian], — 1979. The stratigraphy, tectonics and petroleum geology of the Kopet-Dagh region, northern Iran. Unpublished Ph.D. thesis. Imperial College, London. ager, d. v. 1956. The British Liassic Rhynchonellidae. Part 1 Monograph of the Palaeontographical Society , 110 (476), i-xxvi, 1-50, pis 1-4. — 1965. The adaptation of Mesozoic brachiopods to different environments. Palaeogeography , Palaeo- climatology, Palaeoecology , 1, 143-172. 1969. Alleged sexual dimorphism in Mesozoic brachiopods. 33-36. In westermann, g. e. g. (ed ). Sexual dimorphism in fossil Metazoa and taxonomic implications. International Union Geological Science Series A, No. 1. — 1971. The brachiopods of the erratic blocks of Spilsby Sandstone in Norfolk and Suffolk. Proceedings of the Geologists' Association , 82, 393-402. — 1988. Mesozoic Turkey as part of Europe. 241-245. In audley-charles, m. g. and hallam, a. (eds). Gondwana and Tethys. Geological Society, London, Special Publication No. 37, 317 pp. — childs, a. and pearson, d. a. b. 1972. The evolution of the Mesozoic Rhynchonellida. Geobios , 5, 157-233. — and evamy, b. d. 1963. The geology of the Southern French Jura. Proceedings of the Geologists' Association , 74, 325-356. barczyk, W. 1969. Upper Jurassic terebratulids from the Mesozoic border of the Holy Cross Mountains in Poland. Prace Muzeum Ziemi Warszawa , 14, 1-82. 362 PALAEONTOLOGY, VOLUME 40 berberian, m. and king, G. c. p. 1981. Towards a palaeogeography and tectonic evolution of Iran. Canadian Journal Earth Sciences. 18, 210-265. boullier, A. 1976. Les Terebratulides de EOxfordien du Jura et de la Bordure sud du Bassin de Paris. Annales Scientifiques de P Universite de Besangon. Geologie, 3e Serie , fascicule 27, 457 pp., 25 pis. buckman, s. s. 1918. The Brachiopoda of the Namyau Beds, northern Shan States, Burma. Palaeonto/ogia Indica , New Series , 3, Memoir 2, 254 pp. Childs, a. 1969. Upper Jurassic rhynchonellid brachiopods from northwestern Europe. Bulletin of the British Museum (Natural History ), Supplement 6, 1-119, 12 pis. dagys, a. s. 1968. [Jurassic and Lower Cretaceous brachiopods from northern Siberia.] Akademiya Nauk SSSR. Moscow , 41, 1-126, pis 1-10. [In Russian], gray, j. e. 1840. Synopsis of the contents of the British Museum. 42nd edition. London, 370 pp. — 1848. On the arrangement of the Brachiopoda. Annals and Magazine of Natural History, (2), 2, 435—440. kalantari, A. 1969. Foraminifera from the Middle Jurassic-Cretaceous successions of Kopet-Dagh region (NE Iran). Publication of the NIOC Geological Laboratories , 3, 1-298. — 1987. Biofacies map of Kopet-Dagh region: unpublished map. NIOC Exploration and Production, Teheran, 1 sheet. kuhn, o. 1949. Lehrbuch der Paldozoologie. Stuttgart. v + 326 pp. kziazkiewicz, m. 1974. Contribution a l'etude de la faune du Tithonique de Wozniki (Carpathes Polonaise Occidentales). Acta Geologia Polonica , 24, 437-456. leymerie, a. 1846. Statistique geologique et mineralogique du departement de P Aube. J. B. Baillere lib., Paris, 676 pp., 10 pis. madam, m. 1977. A study of the sedimentology, stratigraphy and regional geology of the Jurassic rocks of eastern Kopet-Dagh, NE Iran. Unpublished Ph.D. thesis. Imperial College, London. makridin, b. p. 1955. [Some Jurassic rhynchorenids of the European part of USSR.] Zapiski Geologicheskogo FakuTtet Kharkov State University , 12, 81-91. [In Russian]. I960. [Order Rhynchonellida and Order Terebratulida]. 239-257, pis 43-52 and 286-305, pis 71-75. In orlov, Yu. a. (ed.). Osnovy Paleontologii. Izdatel’stvo Akademi Nauk USSR, Moscow, 343 pp., 75 pis. [In Russian]. — 1964. [ Brachiopods from the Jurassic of the Russian platform and some adjoining regions ]. Nedra, Moscow, 339 pp., 25 pis. [In Russian]. moussavi-harami, r. 1986. Neocomian (Lower Cretaceous) continental sedimentation in eastern Kopet-Dagh basin in NE Iran. Abstracts of the 12th International Sedimentological Congress , Canberra , Australia, 220 pp. — and brenner, R. l. 1992. Geohistory analysis and petroleum reservoir characteristics of Lower Cretaceous (Neocamian) sandstones. Eastern Kopet-Dagh Basin, northeastern Iran. Bulletin of the American Association of Petroleum Geologists, 76, 1200-1208. munster, G. 1839. Beitrage zur Petrefactenkunde , 1. Bayreuth, 1-124, pis 1-18. smith, a. G., hurley, a. m. and briden, J. c. 1981. Phanerozoic palaeocontinental world maps. Cambridge University Press, Cambridge, 102 pp. sowerby, J. 1816. The mineral conchology of Great Britain. Sowerby, London. Vol. 2 (1815-18), 1-251, pis 103-203. waagen, w. h. 1883. Salt Range fossils. Part 4(2), Brachiopoda. Memoirs of the Geological Survey of India. Palaeontologica Indica, Series 13, 1, fascicule 2, 391-546, pis 29 — 49. M. H. ADABI Department of Geology University of Tasmania GPO Box 252-79 Hobart, Tasmania 7001 Australia Typescript received 5 May 1993 Revised typescript received 18 October 1993 d. v. ager (deceased) A NEW MITRATE FROM THE LOWER ORDOVICIAN OF SOUTHERN FRANCE by MARCELLO RUTA Abstract. The mitrocystitid mitrate Vizcainocarpus dentiger gen. et sp. nov. originates from the Lower Arenig of the Montagne Noire, France. Its distinctive features are two dorsal areas of polyplated integument, thorn- like projections on the postero-ventral head plates and a leftward-opening mouth. A preliminary cladistic analysis shows that V. dentiger is more derived than Chinianocarpos thorali and is the most primitive mitrocystitid with a differentiated centro-dorsal plate D12 and with the upper lip plate n rigidly incorporated in the skeleton. The aims of this paper are to reconstruct and describe a new genus and species of mitrocystitid mitrate from the south of France, and to determine its systematic position. The new form belongs to the diversified fauna of "stylophoran carpoids’ from the Lower Arenig of the Montagne Noire (Thoral 1935; Ubaghs 1961, 1969, 1983, 1991, 1994; Smith 1988; Cripps 19896; Daley 1992). Five or six other lower Ordovician mitrates are known from this region: Peltocystis cornuta, Chinianocarpos thorali , Lagvnocystis cf. pyramidalis , Ovocarpus moncereti , 0.1 circularis and an undescribed peltocystidan mentioned by Ubaghs (1991, p. 157). Jefferies’ calcichordate theory is followed herein. According to this theory, mitrates are primitive members of the three chordate subphyla Acraniata, Tunicata and Craniata. The mitrate described in this paper is a mitrocystitid (sensu Caster 1952). The Mitrocystitida are paraphyletic (Jefferies 1986; Beisswenger 1994) and regarded by Jefferies as stem-group craniates. The reasons for this are explained thoroughly elsewhere (Jefferies 1967, 1968, 1986; Jefferies and Lewis 1978; Cripps 1990; Beisswenger 1994). Mitrates are covered with calcitic plates of echinoderm type and consist of an anterior, massive head, and a posterior, articulated tail. The internal anatomy of the head can be compared, in Jefferies’ view, with that of such primitive living chordates as amphioxus and the tunicates. In particular, the mitrocystitid mitrates would share with modern craniates a lateral line system and dorsal, touch-sensory branches of the trigeminal nerves. MATERIALS AND METHODS Material , horizon and locality. A broken siliceous nodule showing the external aspect of the dorsal and ventral head skeleton, and part of the tail of a single individual, here designated as the holotype (specimens IPM-B 49101a and IPM-B 49101b; Institut de Paleontologie, Museum National d’Histoire Naturelle, Paris). The internal anatomy is unknown. The approximate position of the collecting site locality is shown in Text-figure 1 (reference map for France: IGN, 1:25000; Labastide-Rouayroux 2444-Ouest; co-ordinates: x = 623, 370, y — 3120, 150). The fossils come from the stratigraphical level known as ‘faunizone/’ (Montagne Noire, France). This level consists of dark, micaceous, sandy or silty shales covered with a rusty crust and containing black siliceous nodules of different shapes and sizes (Courtessole et cd. 1982, 1985). The 'faunizone /' occupies most of the formation known as Schistes de Saint-Chinian and is of early Arenig age (Capera et at. 1978). Although a correlation with other localities yielding Arenig sequences is difficult, the trilobite and graptolite associations of "faunizone /’ may indicate an earlier age than that of the earliest IPalaeontology, Vol. 40, Part 2, 1997, pp. 363-383, 2 pis) © The Palaeontological Association 364 PALAEONTOLOGY, VOLUME 40 text-fig. I . Map of the southern side of the Montagne Noire. Stippled areas to the north of Saint-Chinian, Felines and the River Cesse mark the extent of lower Ordovician rocks. The approximate position of the locality where the fossils were collected is indicated by an asterisk. Reference map for France :IGN, 1 :25000; Labastide-Rouayroux 2444-Ouest; coordinates: x = 623,370, y = 3120,150 (redrawn and simplified after Ubaghs 1969). recognizable Arenig beds in Wales. Fortey and Owens (1987) pointed out that a continuous Tremadoc/Arenig sequence in the British Isles is lacking. The lowermost beds of the incomplete Moridunian Stage in Wales incorporate the Merlinia selwynii Biozone. In the Montagne Noire, the level / includes, among the other taxa, the trilobites Asaphelina barroisi , Euloma filacovi, Megistaspis filacovi and Symphysurus angustatus, the graptolites Clonograptus cf. persistens , Dictyonema cf. murrayi and Tetragraptus amii , the articulate brachiopods Orthambonites roquebrunensis and Pleurorthis fascis and the inarticulate brachiopods Spondyglossella spondylifera and Conotreta turricula (Courtessole et al. 1982). The Schistes de Saint-Chinian have yielded well-preserved and articulated skeletons (mainly of echinoderms), regardless of the size and texture of the skeletal elements. This suggests relatively rapid burial of the organisms and limited post-mortem transport. The cornute and mitrate fauna from the Montagne Noire includes, to date, 22 species and represents, together with the radiate echinoderms, \..un des ensembles fauniques les plus diversifies que Ton connaisse’ (Ubaghs 1994, p. 108; see also Smith 1988). However, individual specimens are very rare. According to Ubaghs (1969), this rarity is presumably due to preservational bias and to the fact that only isolated nodules have been collected regularly over the past decades, whereas an accurate study of the nodule-bearing sediments has long been neglected. Methods. The specimens were cleaned in ethanolamine thioglycollate. IPM-B 49101b was partly dissected to expose the partial external mould of the tail. The reconstruction of Vizcainocarpus dentiger (Text-fig. 2) is based on camera lucida drawings of latex casts coated with ammonium chloride. RUTA: ORDOVICIAN MITRATE 365 text-fig. 2. Reconstruction of Vizcainocarpus dentiger gen. et sp. nov. in a, dorsal, b, ventral, c, left lateral, and d, right lateral views. 366 PALAEONTOLOGY, VOLUME 40 text-fig. 3. Plate nomenclature in Chinianocarpos thorali and Vizcainocarpus dentiger. a, reconstruction of the head of C. thorali in dorsal aspect (redrawn after Jefferies 1986). B, reconstruction of the head of V. dentiger in dorsal aspect. Drawings not to scale. SYSTEMATIC PALAEONTOLOGY Superphylum deuterostomia Grobben, 1908 (Stem group of the Craniata?) Family Incertae sedis Genus vizcainocarpus gen. nov. Derivation of name. After M. Daniel Vicai'no of Carcassonne, France, who discovered the fossil described herein, for his outstanding contributions to the palaeontology of the Montagne Noire. The suffix -carpus means a fruit (Greek, karpos). Type species. Vizcainocarpus dentiger sp. nov., the only species known. Diagnosis. Two dorsal areas of polyplated integument; centro-dorsal plate D12 in contact posteriorly with a process projecting from plate i and anteriorly with plates c, d and n; mouth opening leftward and limited ventrally by oral platelets of different shapes and sizes; anterior half of ventral head skeleton with irregular plates; posterior half consisting of hexagonal plates with a conical process bending forward ; stereom texture of the marginal, dorsal and ventral integument plates highly porous; peripheral external surface of plates e and 6 , external surface of plate n and of the oral platelets with coarse texture; ridges of fibrillar stereom visible along the dorsal margins of the styloid blades and the hind-tail ossicles; lateral line opening surrounded by a semicircular thickening and close to the posterior margin of plate e. RUTA: ORDOVICIAN MITRATE 367 Vizcainocarpus dentiger sp. nov. Plates 1-2; Text-figures 2-8 Derivation of name. The specific name alludes to the tooth-like processes on the postero-ventral head plates (Latin, dentiger = tooth-bearer). Diagnosis. As for the genus, by monotypy. ANATOMICAL DESCRIPTION Head The plate nomenclature implies homology with like-named plates in cornutes such as Reticulocarpos hanusi and Prokopicystis mergli, and in mitrates such as Barrandeocarpus norvegicus and Mitrocystella incipiens (Jefferies and Prokop 1972; Craske and Jefferies 1989; Cripps 1989m 1991 ; Cripps and Daley 1994). The head is about 6 mm long and 4 mm across at its widest point. V. dentiger is, therefore, significantly smaller than most other known mitrates, its size being comparable to that of such ‘ dwarf' cornutes as Beryllia miranda, Domfrontia pissotensis , Nanocarpus dolambii and Prokopicystis mergli (Ubaghs 1991; Cripps and Daley 1994). It is not known whether the small size of V. dentiger is related to its ontogenetic age or is a specific feature of this mitrate. The head is roughly pyriform in outline. Its left side is less convex externally than its right side. The dorsal surface is flat, whereas the ventral surface is gently convex. The height of the head is greatest near the tail junction and decreases uniformly towards the mouth opening. The most anterior part of the dorsal skeleton and the left anterior angle of the head are poorly preserved. Four marginal plates on the right and two marginal plates on the left are visible. A large, asymmetrical plate n lies anterior to the left integument area. Plates h and i are the largest dorsal plates, whereas e and 6 are the largest ventral plates. Dorsal skeleton. The dorsal skeleton consists of nine marginal plates, a centro-dorsal plate D12 and two areas of polyplated integument, one on the left and one on the right of D12 (Text-figs 2a, 3b; PI. 1, fig. 1 ; PI. 2, figs 1-2). The posterior and lateral marginal plates show a flat dorsal and a curved ventro-lateral part. Plates h and i build the posterior third of the dorsal skeleton. They differ in shape and size, and are wider posteriorly than anteriorly. The suture between them bends gently rightward. The postero-dorsal margin of both plates is convex rearward and bears seven or eight shallow ‘ teeth ’. Each tooth has a longer, gently sloping median side and a shorter, steep lateral side (Text-fig. 2a; PI. 1, fig. 1 ; PI. 2, fig. 2). Plates h and i are in contact with the rearmost lateral marginal plates of the right and left side, respectively, along two antero-laterally concave sutures (Text-fig. 2c-d; PI. 1, fig- 1). The anterior margin of h is slightly concave forward. The anterior margin of i carries a stout process directed anteriorly and rightward. The left margin of this process is almost straight, whereas its right margin is concave. The process is sutured with the posterior margin of D12 (PI. I, fig. 1 ; PI. 2, fig. 2). Two shallow, peripheral grooves run along the posterior margins of h and i without changing their width and almost reach the dorsal mid-line (Text-figs 2a, 3b). Their position and extent are marked by an abrupt change in the histology of the calcitic skeleton near the posterior margin of the head (PI. 1, fig. 1; PI 2, fig. 2). The peripheral grooves straddle the sutures formed by h and i with the rearmost right (0 and left (al) marginal plates, respectively, fiere, they turn downward and continue as vertical, shallow depressions along the lateral head walls just behind the vertical projections of the i/al and h/f sutures (Text-fig. 2c-d). Most of the dorsal external surface of h and i shows regularly spaced, sub-circular, stereom pores delimited by thick trabeculae (Text-fig. 4; PI. 2, fig. 2). Near the anterior, lateral and median margins of h and i and on the anterior process of i the pores are slightly smaller. At the level of the peripheral grooves they become slit- like and irregular in outline and the trabeculae which delimit them are twisted. Near the median end of each peripheral groove, the pores and trabeculae run obliquely rearward and laterally. Near the lateral ends of both grooves, most trabeculae and pores form a right angle with the main body axis (PI. 2, fig. 2). The downward projections of h and i form most of the posterior head surface (Text-fig. 5; PI. 1, fig. 2) and slope slightly forward and downward. Their external surface shows a compact stereom. Sinuous ridges and irregular striations are also visible along their dorsal margins. Four marginal plates frame the right side of the head (Text-figs 2a, 3b; PI, 1, fig. 1) and are identified as f, e, d, and c on the basis of a comparison with Chinianocarpos thorali (Text-fig. 3a; see also Cripps 1990). Plate f has a convex lateral margin and is slightly wider posteriorly than anteriorly. Its postero-lateral angle bears 368 PALAEONTOLOGY, VOLUME 40 text-fig, 4. Vizcainocarpus dentiger gen. et sp. nov. Reconstruction of the stereom texture of the marginal plates (based mainly on plate i). text-fig. 5. Vizcainocarpus dentiger gen et. sp. nov. Reconstruction of the posterior aspect of the head. a notch corresponding to the point where the most anterior part of the right peripheral groove straddles the posterior end of the h/f suture and bends downward (PI. 1, fig. 1). Plate e is sub-trapezoidal and occupies the central part of the right margin of the head (PI. 1, fig. 1). Plate d is pentagonal and is in contact with e posteriorly, c anteriorly and the most anterior part of D12 medially (PI. 1, fig. 1 ; PI. 2, fig. 1 ). The suture between D12 and d runs almost parallel to the main body axis. Plate c, the smallest dorsal marginal element, lies just right of the dorsal mid-line. Its posterior third is inserted between D12 and d. Its anterior third lies right of plate n (PI. 1, fig. 1 ; PI. 2, figs 1, 5). On the left side of the head two marginal plates are visible (Text-figs 2a, 3b; PI 1, fig. 1). The anterior plate, poorly preserved and best observed in ventral view (PI. 1, fig. 2), occupies the left anterior angle of the head and perhaps corresponds to plate b in C. thorali (Text-fig. 3a). A single, elongate plate frames most of the left side of the head posterior to b. This plate, labelled as al, is gently convex externally. Like f, al shows in dorsal view a notch near its postero-lateral angle, marking the position of the most anterior part of the left peripheral groove (Text-figs 2a, 3b; PI. 1, fig. 1). Plate al corresponds in position to a and I in C. thorali , but it is difficult to say whether it is homologous with one or the other of these elements or whether it results from their fusion. EXPLANATION OF PLATE 1 Figs 1-4. Vizcainocarpus dentiger gen. et sp. nov.; Level f Schistes de Saint-Chinian (Lower Arenig); Cassagnoles, Montagne Noire, France; latex casts. 1, IPM-B 49101a; dorsal head skeleton; x 14. 2-4, IPM- B 49101b. 2, ventral head skeleton; x 14. 3, fore-tail rings and styloid blades in left lateral view; x 40. 4, hind tail in left lateral view; x 40. PLATE 1 RUTA, Vizcainocarpus 370 PALAEONTOLOGY, VOLUME 40 text-fig. 6. Vizcainocarpus dentiger gen. et sp. nov. Camera lucida drawing of the most anterior part of the ventral head skeleton showing the lower lip plates arranged along a slanting line; x 20. An isolated skeletal fragment is visible both dorsally and ventrally in front of the left anterior angle of the head (PI. 1, figs 1-2). The fragment consists of two sloping surfaces. The highly porous stereom texture of one of these surfaces is very similar to that of the dorsal surface of h and i. The fragment is perhaps part of a posterior marginal element of a different individual, but its identification is uncertain. The stereom texture of all marginal plates except n consists of larger pores in a central position, and smaller pores towards their median dorsal margins and near the sutures with adjacent plates. The upper lip plate n is asymmetrical in outline and comparable in relative size, shape and position to its namesake in C. thorali (Text-fig. 3 ; PI. 1, fig. 1). Although the preservation is poor, its posterior margin is partly visible in front of the left dorsal integument area. Its anterior margin runs from left of c to right of b and is anteriorly convex. The n/c suture is represented by a faint sinuous line (PI. 1, fig. 1 ; PI. 2, fig. 1). In C. thorali , a short, narrow area of integument is present between n and c (Text-fig. 3a). This area consists of two or three small plates and projects anteriorly from the integument comprised between the left marginal plates and the anterior half of the oblique ridge complex (Jefferies and Prokop 1972, pi. 7, fig. a; Jefferies 1986, p. 291 ; Cripps 1990, p. 37). In V. dentiger , n is in contact with D12 along the most anterior part of the latter (PI. 1, fig. 1; PI. 2, fig. 1). D12 in V. dentiger can be readily homologized with its namesake in other mitrocystitid mitrates on the basis of its position. This plate runs obliquely from anterior right to posterior left, dividing the dorsal skeleton into two subequal fields. D12 is comparatively narrower in V. dentiger than in all the other mitrates in which this element is differentiated as a separate plate, and consists of two parts. The anterior part is rectangular in outline and is in contact with c, d and n. The posterior part is narrow and elongate and is in contact with the anterior process of i along a short suture. Although D12 and i appear as separate elements (PI. 1, fig. 1; PI. 2, fig. 2), this is probably the result of mechanical displacement and not of breakage, as the posterior end of D12 and the anterior end of the process of i show articulation facets along which these plates were originally in contact. In C. thorali , D12 is not a separate plate and d sends out a median flange corresponding to the anterior half of the oblique ridge complex. This flange is sutured with an extension of plate i directed anteriorly and rightward (Text-fig. 3a). The stereom texture of D12 in V. dentiger is similar to that of the marginal elements. Most of the centro- dorsal surface of this plate bears widely spaced pores. Along its margins the stereom texture is slightly coarser, in that the pores are similar and more irregular. The two areas of polyplated integument separated by D12 differ in size and shape (PI. 1, fig. 1). The left area is better preserved, as a small number of plates have maintained their original position. The integument plates are thin and very irregular in shape, although in some cases this may be the result of breakage, and their stereom texture is retiform. Some of the plates in contact with the left margin of D12 are elongate. The left integument area is delimited on the left by the median margins of al and b, posteriorly by the left margin of the process of i, on the right by the left margin of D12 and anteriorly by the posterior margin of n. The right integument area is delimited on the left by the right margin of D12 and by the right margin of the process of i, posteriorly by the anterior margin of h, on the right by the median margins of e and f, and anteriorly by the postero-median margins of d. The dorsal skeleton does not show other openings. Ventral skeleton. The ventral skeleton includes the oral platelets, the plates of the head floor, and the postero- ventral plates e and 6. The downward bending portions of the dorsal marginal elements are also visible ventrally (Text-figs 2b, 6-8; PI. 1. fig. 2; PI. 2, figs 3-5). Five oral platelets arranged along a slanting line are visible. The right angle of the mouth opening is anterior to the left one (Text-fig. 6; PI. 2, fig. 5). Perhaps one or two other platelets were present in life near the left angle of the mouth opening, but the preservation is too poor to allow the exact number of these elements to RUTA: ORDOVICIAN MITRATE 371 be detected. The platelets are roughly equal in length, but differ in width. The two rightmost platelets are narrow, rectangular elements about half as wide as long. The third platelet is slightly wider, whereas the fourth and fifth platelet are about twice as wide as the first two and are sub-trapezoidal. The external surface of the platelets is rough in texture. The platelets are flexibly articulated with the most anterior head floor plates (PI. 1 , fig. 2; PI. 2, figs 3, 5). The integument area lying just behind the lower lip probably corresponded to the buccal cavity in life. The remaining ventral head plates can be divided into two groups : a smaller, anterior group of irregular elements and a larger, posterior group of polygonal plates. The plates of the anterior group do not seem to overlap each other nor do they overlap the more anterior, smaller elements framing the posterior margin of the oral platelets. The plates of the anterior group have irregular shape and arrangement. Those which are closer to the lateral margins of the head are slightly narrower than those which occupy a more median position (PI. 1, fig. 2; PI. 2, fig. 3). The plates of the posterior group form three transverse rows (PI. 1, fig. 2; PI. 2, fig. 3). They are roughly hexagonal and possess a stout, conical process inclined forward (Text-fig. 7). The size of the processes is text-fig. 7. Vizcainocarpus dentiger gen. et sp. nov. Reconstruction of a postero- ventral head plate in a, external and b, left lateral views. A B apparently related to that of the plates bearing them. On the smaller, lateral plates of each of the three transverse rows the processes are knob-like and have an almost symmetrical outline in lateral view. The distal ends of these processes are formed by the fusion of a number of trabeculae. It is difficult to reconstruct the number of elements in each row, especially near the lateral margins of the head, where the plates are considerably displaced. In the first row, four hexagonal plates are visible. The two admedian plates are about twice as large as the lateral ones. The left lateral plate is poorly preserved and its process is not clearly visible. The second row consists of five hexagonal plates, all of which carry a robust process. The two admedian plates are as large as the admedian plates of the anterior row. The lateral plates of the second row are smaller than the lateral plates of the first row. The mid-ventral plate is the largest polygonal element of the postero- ventral skeleton. In the third row, the two admedian plates and the left lateral plate show a minute process. The right lateral plate is poorly preserved. The plates of the third row are not regular in outline and are less well preserved than those of the first two rows. A small element lying in front of the e/9 suture, apparently without a process, is likely to correspond to plate p of other mitrocystitids (PI. 1, fig. 2; PI. 2, fig. 3). Other small, irregular plates are present between the lateral elements of the first and second, and of the second and third, transverse rows, both on the left and on the right side of the head. Their precise arrangement cannot be reconstructed as accurately as that of the process-bearing plates. The ventral plates show a highly porous stereom texture. The pores are generally irregular in shape and of different sizes even on the same plate, and are delimited by laterally compressed trabeculae that are oriented randomly. The posterior ventral plates show one or two circlets of elongate pores on the external surface of the tooth-like processes. 372 PALAEONTOLOGY, VOLUME 40 text-fig. 8. Vizcainocarpus dentiger gen. et sp. nov. Camera lucida drawing of the right postero-lateral angle of the head, showing plates e, h and f. The lateral line opening is visible on plate e; x40. The two largest ventral plates, e and 6 , are roughly trapezoidal and do not show a regular pattern in the distribution of pores and trabeculae (Text-figs 2b, 8; PI. 1, fig. 2; PI. 2, fig. 4). Plates e and 0 contribute to the lower half of the posterior head excavation and meet along a straight mid-ventral suture. Plate 9 slightly overlaps e in IPM-B 49101b. This is almost certainly due to a post-mortem mechanical displacement of the two plates relative to each other, since 9 bears a flat articulation facet along its median margin and is separated from plate al by a gap. A small sub-circular pit, visible at the centre of the posterior margin of e (Text-fig. 8 ; PI. I . fig. 2; PI. 2, fig. 4), represents the opening of the lateral line system, according to Jefferies ( 1986). The remarkable aspect of this RUTA: ORDOVICIAN MITRATE 373 opening in V. dentiger is its marginal position to e. In all other mitrocystitids, the opening is anterior to the posterior margin of e. The external surface of e and 9 is coarsely granular (Text-fig. 8 ; PI. 2, fig. 4). The central area of each of these plates is occupied by twisted trabeculae which delimit irregular hollows. Around this area the trabeculae have a more regular arrangement, and tend to radiate towards the lateral, median and anterior margins of both t and 9. The anterior margins of e and 9 show an irregular fringe of fibrillar stereom, whereas their posterior and postero-median areas are compact. The rounded postero-lateral edges of e and 9 fit into shallow grooves running along the ventral margins of the downward projections of h and i, and form with them a rocking articulation. This presumably allowed e and 9 to move relative to h and i, respectively, in life. A similar articulation between e and h and between 9 and i in Eumityrocystella scivilli was described by Beisswenger (1994, p. 450) as a mechanism to ‘ allow the atrial openings to gape’. In V. dentiger , there are indications of the possible location of the left and right atrial openings where plates e, h and f, and plates 9 , i and al, meet. The postero-lateral angles of the head were particularly flexible, as revealed by the morphology of the plate sutures (Text-fig. 8) in these regions. Plates e and 9 partly overlap the polygonal plates situated in front of them. They also overlap two small, intercalary plates at the postero-lateral angles of the ventral skeleton (Text-figs 2b, 8; PI. 1, fig. 2). The anatomically left intercalary plate is inserted between 9 and al, whereas the right intercalary plate is inserted between s and f. Tail Fore-tail. The fore-tail skeleton (Text-fig. 2c-d; PI. 1, figs 2-3) consists of five or six calcific rings. Each ring is made of four plates. The dorsal plates seem to be flexibly articulated with the left ventral plates, but the nature of this articulation is difficult to reconstruct. No polyplated folds of integument between successive rings have been observed. The stereom of the rings consists of irregular striations oriented perpendicular to the longitudinal axis of the tail. Despite the poor preservation of the fore-tail, it is possible to observe that the rings are partly clenched ventrally without signs of breakage (PI. 1, fig. 2). Mid-tail. The styloid (Text-fig. 2c-d; PI. 1, fig. 3) is a massive element. Its anterior blade has a semicircular outline. Its posterior blade is incomplete, but the preserved parts of its anterior and posterior margins are sharp. The stereom forms a series of fine striations along the dorsal margin of the anterior blade and on the preserved part of the posterior blade. The lateral styloid surfaces are slightly concave outward. Only one pair of plates has been found articulated with the styloid. These are rectangular elements with a coarse histology comparable to that of the hind-tail plates. The mid-tail plates are slightly larger than the hind-tail plates and, like these, they are obliquely oriented with respect to the longitudinal axis of the tail. Hind- tail. Seven or eight articulated, partially preserved ossicles are visible (Text-fig. 2c-d; PI. 1, fig. 4). The first ossicle is taller than and twice as large as the second and comparable in size to the posterior styloid blade. The ossicles are rectangular in lateral view and their posterior and ventral margins are sinuous. The ossicles decrease rapidly in height so that the preserved region of the hind-tail shows a vague morphological tripartition, although not so evident as in Chinianocarpos thorali , Peltocystis cornuta, Chauvelia discoidalis or Eumitrocy Stella savilli (Jefferies 1986; Cripps 1990; Beisswenger 1994). Irregular striations are visible along the dorsalmost parts of the lateral ossicular surfaces. The styloid blades and the tail ossicles seem to have anterior and posterior cutting edges. The ventral hind-tail plates are rectangular in outline and overlap each other antero-posteriorly. It is not possible to reconstruct the nature of the contact between ossicles and plates. The plates are obliquely oriented with respect to the longitudinal axis of the tail so that the two plates of each hind-tail segment would have formed a chevron in life with the apex pointing rearward, as in Chauvelia discoidalis (Cripps 1990). COMPARISONS V. dentiger resembles Chinianocarpos thorali in the outline of the head, in the shape, position and relative size of plate n, and in the presence of two shallow peripheral grooves running near the posterior margins of h and i. Like C. thorali, V. dentiger has four right marginal plates, a process of plate i directed forward and rightward, and a sub-elliptical area of integument left of the oblique ridge complex. 374 PALAEONTOLOGY, VOLUME 40 V. dentiger differs from C. thorali in possessing a differentiated plate D12. The latter presumably derives from plate d, which in C. thorali sends out a projection directed leftward and rearward. This projection carries along its left margin the anterior half of the oblique ridge and contacts the anterior process of i. In C. thorali , this process is sutured with the left margin of h. In V. dentiger , D12 is completely separate from h and delimits in part the right dorsal integument. Unlike C. thorali , V. dentiger shows plate n locked rigidly to the head frame. The presence of a suture between n and b is only vaguely suggested in the available material of V. dentiger , but the n/c suture is clearly visible. Also, V. dentiger has a flexible lower lip, whereas the lower lip is rigid in C. thorali. Plate al in V. dentiger probably corresponds, at least topologically, to plates a and 1 of C. thorali. In the latter species, 1 is the longest of the three lateral marginal plates of the left side of the head. Plate al in V. dentiger frames most of the left margin of the head. It occupies the same position as a and 1 in C. thorali , and its median margin delimits the left side of the left integument area. The presence of two dorsal areas of polyplated integument in V. dentiger is a character shared with Aspidocarpus and Clumvelia. In these forms, D12 contacts both the two most anterior marginals of the right side of the head and plate n (Ubaghs 1979; Cripps 1990). Like Aspidocarpus and Chauvelia , V. dentiger shows no signs of paripores. As in Aspidocarpus, Chauvelia and Ovocarpus , plates e and 0 in V. dentiger overlap small, scale-like elements lying just median to the rearmost lateral marginals. V. dentiger differs from Aspidocarpus and Chauvelia in that it has a smaller number of lateral marginal plates and a more asymmetrical head outline. Also, its mouth faces more distinctly leftward and its oral platelets lack an anterior process. As in most mitrocystitids, the downward projections of h and i in V. dentiger are visible ventrally. Plates e and 6 in V. dentiger do not show postero-ventral transverse keels, whereas in Ovocarpus, Chauvelia and Aspidocarpus , these keels are evident and each of them delimits two areas of the external surface of e and 0 with a different stereomic structure. Plates e and 0 in V. dentiger contribute to a small portion of the posterior head surface, which is formed mainly by the downward projections of plates h and i. The suture between £ and 0 in V. dentiger is relatively long with respect to the size of these two plates. Among the mitrocystitids, this condition is found only in Mitrocystella incipiens. In all other mitrocystitids, the e/6 suture is relatively short, exceptionally so in Aspidocarpus, Chauvelia and Eumitrocy Stella. V. dentiger resembles Ovocarpus in possessing a highly porous stereom on the dorsal side of the head. The stereom of the dorsal and ventral head plates of Aspidocarpus (Ubaghs 1979) and Chauvelia (Cripps 1990) has comparatively smaller pores. V. dentiger differs from Aspidocarpus, Chauvelia and Ovocarpus in that a porous stereom is also present on its marginal plates. Cripps (1989a, 19896) and Cripps and Daley (1994) proposed a progenetic mechanism to explain the small size and the contemporaneous presence of retiform stereom in such mitrate-like cornutes as Bervllia miranda, Domfrontia pissotensis, Prokopicystis mergli and Reticulocarpos hanusi (progenetic dwarfism). The relatively small size of such mitrates as C. thorali and Peltocystis cornuta is generally considered as an inheritance from their cornute ancestors, whereas the absence of retiform stereom is seen as a secondary loss in adult mitrates. Ontogenetic changes in the histology of the calcitic head plates have been observed in Mitrocystites mitra. Young individuals of this species show a typical retiform stereom (Jefferies 1968, pi. 8, fig. 5), sometimes partly replaced by a more compact stereom in the adults (Ubaghs 1967, fig. 328/ 7c). The change in the stereom structure near the sutures of the marginal plates of V. dentiger is comparable to that observed in Reticulocarpos hanusi and Prokopicystis mergli. Jefferies and Prokop EXPLANATION OF PLATE 2 Figs 1-5. Vizcainocarpus dentiger gen. et sp. nov.; Level /, Schistes de Saint-Chinian (Lower Arenig); Cassagnoles, Montagne Noire, France; latex casts. 1-2, IPM-B 49101a; x 26. 1, right antero-lateral region of dorsal head skeleton. 2, detailed aspect of plate i. 3-5, IPM-B 49101b; x 20. 3, detailed aspect of ventral head skeleton. 4, postero-ventral aspect of head. 5, oral platelets. PLATE 2 RUTA, Vizcainocarpus 376 PALAEONTOLOGY, VOLUME 40 (1972, p. 80) interpreted the change in texture near the plate sutures of Reticulocarpos hanusi as a ‘slowing-down of growth towards the end of life’, and suggested (p. 72) that the preserved individuals of this cornute 'were not juvenile, but adult despite their small size’. Small size and presence of retiform stereom on the integument plates also characterize the mitrate-like cornute Nanocarpus dolambii (Ubaghs, 1991) and can perhaps be interpreted as a consequence of heterochronous evolution. Ubaghs (1994) reasoned that the presence of a wide- meshed stereom on the dorsal surface of the mitrate Ovocarpus moncereti is either a primitive or a neotenous feature of this form. On the basis of the available material of V. dentiger , it is difficult to say if this mitrate was characterized by dwarfism. Among mitrates, only Ovocarpusl circularis (Ubaghs, 1994) is comparable in size to V. dentiger and shows a similar retiform stereom. PHYLOGENETIC ANALYSIS The character coding and most of the statistical aspects of the analysis (Text-fig. 9) will be thoroughly discussed elsewhere. In this section, I mainly focus on V. dentiger. Seventeen taxa and 76 characters were chosen (see Appendices 1 and 2). The taxa are : Aspidocarpus bohemicus (Ubaghs, 1979), Ateleocystites guttenbergensis (Kolata and Jollie, 1982), Barrandeocarpus jaekeli (Ubaghs, 1979), B. novegicus (Craske and Jefferies, 1989), Chauvelia discoidalis (Cripps, 1990), Chinianocarpos thorali (Ubaghs, 1961), Domfrontia pissotensis (Cripps and Daley, 1994), Eumitrocy Stella savilli (Beisswenger, 1994), Lagynocystis pyramidalis (Jaekel, 1918), Mitrocystella barrandei (Jaekel, 1901), M. incipiens (Jaekel, 1901), Mitrocystites mitra (Barrande, 1887), Ovocarpus moncereti (Ubaghs, 1994), Peltocyst is cornua (Thoral, 1935), Placocystites forbesianus (de Koninck, 1869), Prokopicystis mergli (Cripps, 1989a), Vizcainocarpus dentiger gen. et sp. nov. The multistate coding in some taxa expresses uncertainty and is represented by states 1 or 2 in almost all cases, the only exception being character number 6 (0 or 2) in Ovocarpus moncereti. All characters were left unweighted and unordered. The cornutes Domfrontia pissotensis and Prokopicystis mergli and the mitrates Lagynocystis pyramidalis and Peltocystis cornuta were used as outgroups to polarize the characters. The mitrates Ateleocystites guttenbergensis and Placocystites forbesianus were chosen as representatives of the anomalocystitids. The data were processed with PAUP 3.1.1 (Swofford 1993). The branch-and-bound search found one parsimonious tree (length = 195; Cl = 0-687; RI = 0-777). The character-state optimization used was the accelerated transformation (ACCTRAN), which maximizes synapomorphies at each node and emphasizes reversals. The tree was rooted so as to make the outgroup paraphyletic with respect to the ingroup. The position of V. dentiger (node C) is supported by ten state changes relative to characters nos 23 (0-*l; Cl = 1-000), 29 (2-*l; Cl = 1 000), 30 (2^0; Cl = 0-667), 36 (l-*2; Cl = 1 000), 44 (1-^2; Cl = 0-500), 46 (0 — s- 1 ; Cl = 1 000), 49 (2 -> 0 ; Cl = 0-667), 55 (2 ^ 1 ; Cl = 0-400), 63 (2->l; Cl = 1 000), 71 (0->l; Cl = 1 000). V. dentiger shows a multistate coding for character numbers 28, 51, 58, 61 and 76, concerning the position of plate n with respect to the marginal plates, the nature of the contact between the dorsal and ventral fore-tail plates, and the morphology of the first hind-tail ossicle. Some phylogenetic deductions from the cladogram are possible. The differentiation of D12 in V. dentiger led to the appearance of flexible integument areas over the left and the right pharynx. The widening of D12 reduced the relative extension of these areas (e.g. in Aspidocarpus and Chauvelia). Further reduction in the number of plates on both sides of D12 characterizes Mitrocystites mitra , in which the dorsal skeleton is completely rigid. In all other mitrocystitids and in the anomalocystitids D12 is expanded transversely and occupies most of the centro-dorsal skeleton. The changes outlined above are perhaps linked with a number of modifications that occurred inside the mitrate head. These modifications would probably have mainly affected the pharynx. A particular feature of the mitrate head, the oblique groove, is important in this respect. According to Jefferies (1986), the oblique groove of mitrates separated the primary, left pharynx with the overlying left anterior coelom from the right anterior coelom. This groove runs obliquely across the RUTA: ORDOVICIAN MITRATE 3 o 6 TO •C o -Q i§ s- 8 •§ & •52 6 o TO "6 •5 0) 3 TO ■C O 6 S & p ■5 S> 1- p ■8 c 8 TO -O I P I § I I P 5 dj CD (D •55< w 3 e- 8 o ■p c 8 TO CQ TO 8 O) I o c TO 3 8- 8 o ■p c 8 TO CQ .TO TO C CD •O C 8 O) TO TO I O TO TO TO 3 C .TO TO TO ■e >2 TO TO I P O to 377 text-fig. 9. Interrelationships of the Mitrocystitidae. Numbers on the tree represent the synapomorphies supporting a particular node or the changes characterizing a particular branch. Only nodes C, D and E. which are relevant to the position of Vizcainocarpus dentiger , are described. dorsal head steinkern and corresponds to an intercameral oblique ridge on the ventral surface of the dorsal skeleton. In Chinianocarpos thorali , the part of the dorsal head steinkern corresponding in position to the left pharynx is bigger than that corresponding to the anterior coeolom plus the right pharynx. Also, the oblique ridge is straight and lies right of the patch of dorsal integument overlying the left RUTA ORDOVICIAN MITRATE 378 PALAEONTOLOGY, VOLUME 40 pharynx (Jefferies and Prokop 1972, fig. 30, 3; pi. 14, fig. lb). Although internal head anatomy is not known in V. dentiger , it is possible to deduce that the anterior half of the oblique ridge lay more to the left than in C. thorali. The leftward displacement of the oblique ridge is more evident in Aspidocarpus and Chauvelia, which show a relatively wide, right dorsal flexible integument area. The position of the oblique ridge in these forms perhaps results from an enlargement of the second right pharynx, which, in turn, was accompanied by an increased pumping activity. In cladistically more derived mitrocystitids, it is possible to notice a general trend towards an increase in head size (e.g. Mitrocystites mitra and Mitrocystella incipiens). Perhaps as a consequence of the need for major stability during locomotion, the head roof became rigid. This rigidity was achieved through the widening of the dorsal head plates, which became less numerous. The latter process culminated with the presence of only two or three centro-dorsal plates in such forms as Mitrocystella spp., Eumitrocy Stella savilli , Barrandeocarpus spp. and the Anomalocystitida. A rigid ventral skeleton and a bilaterally symmetrical head (a condition exemplified by most Anomalocystitida) might also have been selected as a response to increased efficiency in locomotion. Acknowledgements. I thank Dr R. P. S. Jefferies (Palaeontology Department, The Natural History Museum, London) for suggesting and supervising this paper and M. Daniel Vizcaino of Carcassone, France, for collecting the fossil material and making it available for study. Dr A. R. Milner (Birkbeck College, University of London) and Professor S. K. Donovan (University of West Indies, Jamaica) read an early draft of the manuscript and offered invaluable suggestions to improve its style and content. This paper benefited greatly from general discussions on stratigraphy and cladistic methodology with Drs P. L. Forey, R. A. Fortey, A. B. Smith and P. D. Taylor, all of the Palaeontology Department of The Natural History Museum, London. Dr P. E. J. Daley, formerly of the same department, processed the data matrix independently with the program Hennig86. 1 thank Mr Phil Crabb and Mr Harry Taylor (Photographic Unit) and the staff of the palaeontology laboratory of The Natural History Museum, London, for the photographs and for the preparation of the specimens. I am particularly grateful to Drs L. R. M. Cocks, Keeper, and S. J. Culver, Associate Keeper of the Palaeontology Department of The Natural History Museum, London, for their kind hospitality in their institution and to Dr P. E. Ahlberg, Miss S. Evans, Mr D. N. Lewis, Dr A. C. Milner, Mr N. Monks, Mr K. J. Tilbrook and especially Mr B. Lefebvre and Miss M. Marti Mus for lively discussions. This paper forms part of a Ph.D. project on the evolutionary history of the mitrates carried out by the author at the University of London (Birkbeck College). The author gratefully acknowledges the receipt of a grant from the European Community (Training and Mobility of Researchers Programme). REFERENCES barrande, J. 1887. Systeme silurien du centre de la Boheme , volume 7. Classe des Echinodermes. Ordre des cvstidees. Rivnac, Prague, 233 pp. beisswenger, M. 1994. A calcichordate interpretation of the new mitrate Eumitrocy Stella savilli from the Ordovician of Morocco. Paldontologische Zeitschrift , 68, 443^162. capera, j. c., courtessole, r. and pillet, j. 1978. Contribution a l’etude de l’Ordovicien inferieur de la Montagne Noire (France Meridionale). Biostratigraphie et revision des Agnostida. Annales de la Societe Geologique du Nord. 98, 67-88. caster, K. e. 1952. Concerning Enoploura of the Upper Ordovician and its relation to other carpoid Echinodermata. Bulletins of American Paleontology , 34, 1-47. courtessole, r., pillet, j. and vizcaino, d. 1982. Aperqu stratigraphique. 7-22. In babin, c., courtessole, r., melou, m., pillet, j., vizcaino, d. and yochelson, e. l. (eds). Brachiopodes (Articules) et Mollusques (Bivalves, Rostroconches, Monoplacophores, Gastropodes) de FOrdovicien inferieur (Tremadocien- Aremgien) de la Montagne Noire (France Meridionale). Memoire de la Societe d Etudes Scientifiques de I Aude, Carcassonne , 63 pp. - 1985. Stratigraphie. 7-36. In courtessole, r., pillet, j., vizcaino, d. and eschard, r. (eds). Etude biostratigraphique et sedimentologique des formations arenacees de FArenigien du Saint-Chinianais oriental (Herault) versant sud de la Montagne Noire (France Meridionale). Memoire de la Societe d Etudes Scientifique de I Aude, Carcassonne, 57 pp. craske, a. j. and Jefferies, r. p. s. 1989. A new mitrate from the Upper Ordovician of Norway, and a new approach to subdividing a plesion. Palaeontology, 32, 69-99. RUTA: ORDOVICIAN MITRATE 379 cripps, A. p. 1989a. A new stem-group chordate from the Llandeilo of Czechoslovakia and the cornute-mitrate transition. Zoological Journal of the Linnean Society , 96. 49-85. — 19896. A new genus of stem chordate (Cornuta) from the Lower and Middle Ordovician of Czechoslovakia and the origin of the bilateral symmetry in the chordates. Geobios , 22, 215-245. 1990. A new stem craniate from the Ordovician of Morocco and the search for the sister group of the craniata. Zoological Journal of the Linnean Society , 100, 27-71. 1991. A cladistic analysis of the cornutes (stem-chordates). Zoological Journal of the Linnean Society. 102, 333-366. — and daley, p. e. j. 1994. Two cornutes from the Middle Ordovician (Llandeilo) of Normandy, France, and a reinterpretation of Milonicystis kerfornei. Palaeontographica , Abteilung A , 232, 99-132. daley, p. e. j. 1992. Two new cornutes from the Lower Ordovician of Shropshire and Southern France. Palaeontology , 35, 127-148. fortey, R. A. and OWENS, R. M. 1987. The Arenig Series in South Wales. Bulletin of the British Museum ( Natural History ), Geology Series , 41, 69-307. grobben, K. 1908. Die systematische Einteilung des Tierreiches. Verhandlungen der Zoologisch- Botanischen Gesellschaft in Wien , 58, 491-51 I jaekel, o. 1901. Uber Carpoideen, eine neue Klasse von Pelmatozoen. Zeitschrift der Deutschen Geologischen Gesellschaft, 52, 661-677. 1918. Phylogenie und System der Pelmatozoen. Paldontologische Zeitschrift, 3. 1-128. jefferies, r. p. s. 1967. Some fossil chordates with echinoderm affinities. 163-208. In millot, n. ( ed . ) . Echinoderm biology. Academic Press, London, 240 pp. 1968. The subphylum Calcichordata (Jefferies 1967) - primitive fossil chordates with echinoderm affinities. Bulletin of the British Museum (Natural History ), Geology Series, 16, 243-339. — 1986. The ancestry of the vertebrates. British Museum (Natural History), London, 376 pp. — and lewis, d. n. 1978. The English Silurian fossil Placocvstites forbesianus and the ancestry of the vertebrates. Philosophical Transactions of the Royal Society of London, Series B, 282, 205-323. — and prokop, r. j. 1972. A new calcichordate from the Ordovician of Bohemia and its anatomy, adaptations and relationships. Biological Journal of the Linnean Society, 4, 69-115. kolata, d. r. and jollie, M. 1982. Anomalocystitid mitrates (Stylophora, Echinodermata) from the Champlainian (Middle Ordovician) Guttenberg Formation of the Upper Mississippi Valley Region. Journal of Paleontology, 56, 531-565. koninck, m. l. de 1869. Sur quelques echinodermes remarquables des terrains paleozoiques. Bulletin de F Academie Royale des Sciences Belgique, 28, 544-552. smith, a. b. 1988. Patterns of diversification and extinctions in Early Palaeozoic echinoderms. Palaeontology, 31, 799-828. swofford, d. l. 1993. PAUP Phylogenetic Analysis Using Parsimony, Version 3.1.1. Illinois Natural History Survey, Champaign, Illinois, 257 pp. thoral, m. 1935. Contribution a 1' etude paleontologique de FOrdovicien inferieur de la Montague Noire et revision sommaire de la fame cambrienne de la Montague Noire. Imprimerie de la Charite, Montpellier, 363 pp. ubaghs, G. 1961. Un echinoderme nouveau de la classe des Carpoi'des dans POrdovicien inferieur du departement de FHerault (France). Compte Rendu Hebdomadaire des Seances de F Academie des Sciences, Paris, 253, 2565-2567. 1967. Stylophora. S496-565. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part S. Echinodermata 1(2). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 352 pp. — 1969. Les echinodermes carpoides de I' Ordovicien inferieur de la Montague Noire (France). Cahiers de Paleontologie. Editions du Centre National de la Recherche Scientifique, Paris, I 12 pp. 1979. Trois Mitrata (Echinodermata: Stylophora) nouveaux de FOrdovicien de Tchecoslovaquie. Paldontologische Zeitschrift, 53, 98-119. 1983. Echinodermata. Notes sur les echinodermes de FOrdovicien inferieur de la Montagne Noire (France). 33-55. In courtessole, r., marek, l., pillet, j., ubaghs, j. and vizca'ino, d. (eds). Calymenina, Echinodermata et Hyolitha de FOrdovicien de la Motagne Noire (France Meridionale). Memoire de la Societe d Etudes Scientifiques de FAude, Carcassonne , 62 pp. 1991. Deux Stylophora (Homalozoa Echinodermata) nouveaux pour FOrdovicien inferieur de la Montagne Noire (France Meridionale). Paldontologische Zeitschrift , 65, 157-171. 380 PALAEONTOLOGY, VOLUME 40 1994. Echinodermes nouveaux (Stylophora, Eocrinoidea) de l'Ordovicien inferieur de la Montagne Noire (France). Annales de Paleontologies Invertebres, 80, 107-141. MARCELLO RUTA Department of Biology Birkbeck College Malet Street, London WC1E 7HK, UK and Department of Palaeontology Typescript received 26 January 1996 The Natural History Museum Revised typescript received 29 July 1996 Cromwell Road, London SW7 5BD, UK APPENDIX 1 Description of the characters used in the phylogenetic analysis. For each character the different states are indicated by numbers in brackets. A discussion of the characters is given in the text. 1. Presence (1) or absence (0) of oral spines. 2. Postero-median extensions of peripheral grooves absent (0) or present and close to the postero-dorsal margins of the head (1) or not (2). 3. Antero-lateral extensions of peripheral grooves absent (0) or present and not extending (1) or extending (2) onto the second posterior marginal plate on both sides of the head. 4. Lateripores absent (0) or present and anterior to (1) or at the transverse level (2) of the postero-dorsal end of the h/i suture. 5. Presence (1) or absence (0) of paripores. 6. h/i suture straight and aligned with the tail axis (0) or bending rightward (1) or straight but not aligned with the tail axis (2). 7. Posterior margin of both h and i convex posteriorly (0) or with median excavation which occupies less than half (1) or more than half (2) of the maximum width of both h and i or straight (3). 8. Anterior margin of i not lying (0) or lying anterior to anterior margin of h and semicircular in outline (1) or not (2). 9. Projections of h and i visible in ventral view (1) or not (0). 10. h and i not in contact (0) or in contact with the most posterior lateral marginals (1) and with the postero- lateral ends of the sutures thus formed lying anterior (1) or posterior (2) to the postero-dorsal end of the h/i suture. 11. Presence (1) or absence (0) of serrations along the posterior margin of both h and i. 12. Presence (1) or absence (0) of serrations along the external margin of the rearmost right marginal plate. 13. Presence (1) or absence (0) of serrations along the external margin of the rearmost left marginal plate. 14. Presence (1) or absence (0) or dorsal ornament. 15. Outline of the head asymmetrical (1) or symmetrical (0). 16. Mouth opening perpendicular to the longitudinal axis of the tail (1) or not (0). 17. Presence of two (2), three (0), four (1) or live (3) plates on the left side of the head. 18. Presence of three (0), four (1) or live (2) plates on the right side of the head. 19. Presence ( 1 ) or absence (0) of a fringe of striations along the external margins of the lateral marginal plates. 20. Presence (1) or absence (0) of Dll. 21. D10 absent (0) or present and in contact with the most anterior four ( 1 ) or three (2) left marginal plates. 22. Presence ( 1 ) or absence (0) of an oblique ridge complex. 23. D12 in contact with the two most anterior right marginal plates and separated from remaining right marginal plates by an area of polyplated integument (1) or by a single row of plates (2) or in contact with the three most anterior marginal plates and separated from remaining right marginal plates by a single element (3) or none of these conditions (0). 24. Presence ( 1 ) or absence (0) of area of polyplated integument left of oblique ridge complex. 25. D12 contacts the two rearmost left marginal plates posterior to the transverse level of D10 (2) or not (1). 26. Rearmost right marginal plate inserted between the second most posterior right marginal plate and h (1) or not (0). 27. D12 in contact with h (2) or not (1). RUTA: ORDOVICIAN MITRATE 381 28. n absent (0) or present and excluded from (1) or included (2) in the frame of head marginal plates. 29. Dorsal stereom texture of lateral marginal plates divided into a labyrinthic and a retiform region (0) or entirely labyrinthic (1) or compact (2). 30. Stereom texture of dorsal head skeleton retiform (0) or labyrinthic (1) or compact (2). 31. b and c delimit dorso-lateral parts of upper lip (1) or not (0). 32. Anterior margin of e and 9 rounded (1) or polygonal (2). 33. Transverse axis of e and 9 longer (1) or shorter (2) than longitudinal axis. 34. External stereom texture of e and 9 divided into two transverse areas with different histology ( 1 ) or not (2). 35. Posterior margin of e and 9 with median excavation (1) or not (2). 36. Anterior and antero-lateral margins of both s and 9 without (1) or with (2) fringe of striations. 37. e and 9 overlap rearmost lateral marginal plates (1) or scale-like elements comprised between them and the rearmost lateral marginals (2) or form sutures with both the scale-like elements and the rearmost lateral marginals (3) or are sutured with the lateral marginals only (4). 38. Lateral line opening absent (0) or present and sub-circular (1) or groove-like (2) or branched (3). 39. Presence (1) or absence (0) of plate p. 40. Presence ( 1 ) or absence (0) of placocystid plate. 41. Presence (0) or absence (1) of a strut bar. 42. Ventral ornament absent (0) or present and not extending (1) or extending (2) in front of e and 9. 43. Postero-ventral tessellate bar absent (0) or present and with admedian elements separated (1) or not (2) by p. 44. Oral platelets absent (0) or present and framed posteriorly by a semicircular band of small integument plates ( 1 ) or not (2). 45. Anterior margin of oral platelets tooth-like (1) or not (2). 46. Ventral head skeleton with a central region of larger plates with two lateral, narrow longitudinal strips of polyplated integument with numerous small (1) or few elements (2) or none of these conditions (0). 47. All ventral plates arranged in transverse rows (1) or not (0). 48. All ventral plates tessellate (1) or not (0). 49. Stereom texture of ventral head skeleton retiform (0), labyrinthic (1) or compact (2). 50. Presence (1) or absence (0) of ornament on the ventral projections of the lateral marginal plates. 51 Dorsal and ventral fore-tail plates meet along a flexible (1) or rigid (2) suture. 52. Presence (2) or absence (1) of thickening along posterior margin of dorsal fore-tail plates. 53. Presence (1) or absence (0) of spike-shaped plates on dorsal surface of fore tail. 54. Presence of multiple cusps (1) or two blades (2) on the styloid. 55. Anterior styloid blade with rounded profile (1) or not (2). 56. Anterior and posterior margins of posterior styloid blades cutting (1) or posterior surface of posterior blade flat (2). 57. Posterior styloid blade taller than anterior blade (2) or not (1). 58. First hind-tail ossicle with postero-dorsal bearing surface (2) or cutting (1) edges. 59. Ventral mid-tail plates absent (0) or present and separate from (1) or in contact with (2) the styloid. 60. Ventral margins of hind-tail plates meeting at an acute angle (1) or forming a continuous bearing surface (2). 61. Proximal hind-tail ossicles with bearing surfaces (2) or cutting edges (1). 62. Dorsal longitudinal canal of hind-tail absent (0) or present and partially (1) or completely (2) separated from hind-tail lumen. 63. Presence ( 1 ) or absence (0) of fibrillar stereom near dorsal margin of styloid blades and hind-tail ossicles. 64. Presence (1) or absence (0) of specialized olfactory areas. 65. Hypocerebral processes absent (0) or present and delimiting an optic foramen facing ventrally (1) or postero-ventrally (2). 66. Resorption areas of dorsal skeleton absent (0) or present and not forming ( 1 ) or forming (2) extensive fields over the two pharynges. 67. Anterior part of oblique ridge straight (1) or convex rightward (2). 68. Nerves n4 and n5 not differentiated (0) or differentiated and separated distally (2) or not (1). 69. Dorsal head skeleton completely flat (1) or not (0). 70. n subtrapezoidal and with posterior margin longer (1) or shorter (2) than anterior margin. 71. D12 contacts most of the posterior margin of n (2) or its right posterior angle (1) or none of these conditions (0). 382 PALAEONTOLOGY, VOLUME 40 72. Plates b and c fan-like, with transverse axis greater than longitudinal axis and with deeply concave posterior margin ( 1 ) or not (0). 73. Antero-lateral margin of n strongly convex anteriorly, longer than posterior margin and directed from anterior right to posterior left (1) or not (2). 74. First hind-tail ossicle about twice as large as successive ossicle (1) or not (2). 75. p sub-quadrangular and inserted between s and 0 (1) or polygonal (2) or peltate (3). 76. Distal hind-tail ossicles with bearing surfaces (1) or cutting (2) edges. APPENDIX 2 Data matrix. The characters are numbered consecutively on the horizontal axis above the matrix. The symbols used are as follows: ? = unknown character-state; u = uncertainty in the assignment of a character-state expressed as multistate coding. Some statistical aspects of the analysis are explained in the text. Character number 2 3 4 5 6 7 11111111 901234567 1 2 9 0 Taxa Aspidocarpus bohemicus Ateleocystites guttenbergensis Barrandeocarpus jaekeli Barranedocarpus norvegicus Chauvelia discoidalis Chinianocarpos thorali Domfrontia pissotensis Eumitrocystella savilli Lagynocystis pyramidalis Mitrocystella barrandei Mitrocystella incipiens Mitrocystites mitra Ovocarpus moncereti Peltocystis cornuta Placocystites forbesianus Prokopicystis mergli Vizcainocarpus dentiger Character number Taxa Aspidocarpus bohemicus A teleocystites guttenbergensis Barrandeocarpus jaekeli Barrandeocarpus norvegicus Chauvelia discoidalis Chinianocarpos thorali Domfrontia pissotensis Eumitrocystella savilli Lagynocystis pyramidalis Mitrocystella barrandei Mitrocystella incipiens Mitrocystites mitra Ovocarpus moncereti Peltocystis cornuta 0 2 1 10 2 0 0 1 100000200 000000200 000000200 0 2 0 10 10 1 1 0 1 0 0 1 1 0 2 1 0 0 0 0 0 7 7 0 0 000000200 000001220 0 0 0 0 0 1 10 0 0 0 0 0 0 0 10 1 0 2 2 2 1 1 12 1 0 2 1 1 0 u 0 7 1 000001 3 20 1 00000200 000000000 0 1 0 0 0 1 0 2 1 222222222 123456789 0 12 1 10 2 2 1 210020222 2 10 0 2 1 2 2 2 2 1 0 0 1 1 2 2 2 0 11110 12 1 0 10 10 10 12 000000000 2 10 0 2 1 2 0 2 0 10 0 0 1 0 1 2 1 1 0 0 2 1 2 2 2 210020222 2 13 0 10 12 2 7 1 7 7 7 7 7 2 1 0 10 0 0 0 0 1 2 10000001210 2 0 0 0 111110 1 2 0 0 0 10 1 1 10 1 2 0 0 0 1 11110 1 10 0 0 0 0 1 12 1 0 10 1 I 0 0 0 0 I 0 0 00000070000 20000001200 10000000000 2000000320 I 20000001201 2 0 1 10 0 13 2 0 1 1 0 0 0 0 7 7 111 ? 1 1000002100 2 0 0 0 111110 0 00010001 100 1 1000002100 33333333334 01234567890 112 1112 2 110 2 12 2 2 114 0 11 2 12 2 2 2 14 0 1 1 2 12 2 2 1 14 0 1 1 1111112 2 110 2 0 112 111110 00000000000 2 12 1 2 2 14 2 10 2 0 12 2 111 0 10 2 12 12 2 13 2 10 2 12 12 113 3 10 2 12 12 113 2 10 0 7 2 111 2 2 1 1 ? 2 0 2 2 2 1 1 1 0 7 0 RUTA: ORDOVICIAN MITRATE 383 Placocystites forbesianus 2 1 0 0 2 0 2 2 2 2 1 2 2 2 1 1 4 0 1 1 Prokopicystis mergli 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Vizcainocarpus dentiger 0 1 1 1 1 0 1 u 1 0 0 2 1 2 1 2 2 1 1 0 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 5 5 6 Character number Taxa I 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 Aspidocarpus bohemicus 1 0 0 2 1 1 0 0 1 0 u 1 0 2 9 1 ? ? 2 ? A teleocystites guttenbergensis 1 2 2 2 2 0 1 1 2 1 2 2 0 2 2 2 2 2 2 2 Barrandeocarpus jaekeli I 2 2 2 2 0 0 0 2 1 2 2 0 2 2 2 2 2 ? 2 Barrandeocarpus norvegicus 1 2 2 2 2 0 0 1 2 1 2 2 0 2 2 2 2 2 2 2 Chauvelia discoidalis 1 0 0 2 1 2 0 0 1 0 ? ? ? 2 1 1 2 1 2 1 Chinianocarpos thorali 1 0 0 0 0 0 0 0 2 0 0 0 1 2 2 1 2 1 2 1 Domfrontia pissotensis 0 0 0 0 0 0 0 0 0 0 9 ? ? 0 0 0 0 0 0 0 Eumitrocy Stella savilli 1 0 1 2 2 0 0 0 2 0 u 1 0 2 1 2 2 2 2 2 Lagynocystis pyramidalis 1 0 0 2 2 0 0 0 2 0 0 0 ] 1 0 0 0 1 1 1 Mitrocystella barrandei 1 0 1 1 2 0 0 0 2 0 1 1 0 2 2 2 1 2 2 1 Mitrocystella incipiens 1 1 1 1 2 0 0 0 2 1 1 1 0 2 2 2 1 2 2 1 Mitrocystites mitra 1 0 0 1 2 0 0 0 2 0 1 1 0 2 2 2 2 2 2 1 Ovocarpus moncereti 1 ? ? ? ? ? 0 0 ? 0 9 ? ? 2 2 2 2 ? ? ? Peltocvstis cornuta 1 0 0 0 0 0 0 0 2 0 0 1 0 2 1 1 1 1 2 1 Placocystites forbesianus 1 2 2 2 2 0 1 1 2 1 2 2 0 2 2 2 1 2 2 2 Prokopicystis mergli 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Vizcainocarpus dentiger 1 0 0 2 2 1 0 0 0 0 u 1 0 2 1 1 2 u 2 1 6 6 6 6 6 6 6 6 6 7 7 7 7 7 7 7 Character number Taxa 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 Aspidocarpus bohemicus ? ? 1 ? u ? 2 ? 1 1 2 1 2 ? 2 ? Ateleocystites guttenbergensis 2 0 2 1 2 2 2 2 0 2 0 0 2 2 3 2 Barrandeocarpus jaekeli 2 ? 2 9 9 9 9 9 0 2 0 0 2 2 3 2 Barrandeocarpus norvegicus 2 0 2 1 2 2 2 ? 0 2 0 0 2 2 3 2 Chauvelia discoidalis i 1 1 1 2 0 2 2 1 1 2 1 2 1 2 2 Chinianocarpos thorali i 0 2 1 2 0 1 0 1 2 0 0 1 1 2 i Domfrontia pissotensis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Eumitrocy st elia savilli 2 2 2 1 2 2 2 2 0 0 0 0 0 2 2 1 Lagynocystis pyramidalis I 0 2 0 0 0 i 0 0 2 0 0 2 2 1 1 Mitrocystella barrandei 2 2 2 1 2 1 2 ? 0 2 0 0 2 2 2 2 Mitrocystella incipiens 2 2 2 1 2 1 2 2 0 2 0 0 2 2 2 2 Mitrocystites mitra 2 2 2 1 2 1 2 i 0 2 0 1 2 2 2 2 Ovocarpus moncereti ? ? 1 ? ? ? 2 ? 1 2 ? 1 2 ? 2 ? Peltocvstis cornuta 1 0 2 1 1 0 1 0 0 2 0 0 2 2 ? i Placocystites forbesianus 2 0 2 1 2 2 2 2 0 2 2 0 2 2 3 2 Prokopiocystis mergli 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Vizcainocarpus dentiger U ? 1 ? ? ? ? ? 1 2 1 0 1 1 2 U MODE OF LIFE OF THE MIDDLE CAMBRIAN EL DON I O I D LOPHOPHORATE ROTADISCUS by JERZY DZIK, ZHAO YUANLONG and ZHU MAOYAN Abstract. The discs of the alleged 'medusoids' Rotadiscus from the recently discovered Mid Cambrian soft- bodied Kaili fauna in South China are commonly overgrown by shelled epizoans of unknown affinities. Most are attached to the convex side of the disc, near its margin. Towards the disc centre only small shells occur usually, which suggests that their growth was inhibited by an anoxic environment under the disc. The eldonioids were thus sedentary animals, analogous to Recent deep-water thecocyathid corals, with the discs passively lying on the mud surface with the convex side of the disc orientated downward. The assumption that coelenterates, including medusae, are represented among the oldest fossils is questionable for various reasons, both with respect to Cambrian (Dzik 1991) and Vendian (Seilacher 1989) faunas. Moreover, some pyritized fossils from the early Devonian Hiinsruck slate of western Germany that were proposed to represent fossil siphonophorans (Stanley 1986), as well as alleged ctenophorans from that locality (Stanley and Stunner 1983, 1987), are more likely to be poorly preserved shelly or echinoderm fossils (Otto 1994). No convincing evidence is thus available for the presence of medusoid coelenterates prior to the mid Palaeozoic. Some Cambrian fossils, although clearly of non-coelenterate anatomical organization, continue to be called the ‘ medusoids" of that period, ecological analogues of the Recent scyphozoans (Chen et al. 19956; Rigby and Milsom 1996). These are the Eldonioidea (Dzik 1991), Cambrian animals with a U-shaped gut enclosed in a low conical, discoidal body bearing a biramous tentacular apparatus - features shared with some lophophorates and echinoderms. Unlike echinoderms, the eldonioids do not have any mesodermal calcitic skeleton but, instead, a cuticular cover secreted marginally by a kind of mantle (Dzik 1991 ). In some cases this external skeleton, bearing distinct growth lines, was stiff enough to be preserved in coarse sandstones, originating as mass flow sediments (Masiak and Zylihska 1994). In this paper we provide additional evidence to that presented by Dzik (1991) for the sedentary mode of life of the eldonioids. This is offered by new data on a Rotadiscus species occurring abundantly in the Mid Cambrian Kaili fauna (Zhu and Zhao 1995) in Guizhou Province, China. Several hundred Rotadiscus specimens (mostly incomplete) have now been collected; about one- third of them are overgrown with epibionts. The studied specimens are stored at the Department of Geology of the Guizhou Institute of Technology in Guiyang (abbreviated GTB) and the Institute of Paleobiology of the Polish Academy of Sciences in Warsaw (ZPAL). GEOLOGICAL SETTING Rotadiscus is most common within bed 24 in the upper part of the Kaili Formation. This unit is about 23 metres thick and the eldonioid-rich horizons occur in proximity to the first calcareous mudstone bed in the formation (Text-fig. 1 ), the type of rock that characterizes the overlying bed 25 and higher units. The strata are rich in articulated pelagic and benthic trilobites and other excellently preserved fossils. These include abundant uncalcified algal thalli and rare isolated Wiwaxia sclerites. The mudstone is regularly laminated, without bioturbation or preserved infaunal organisms. The general lack of bioturbation suggests low-oxygen conditions near the sediment- water interface. Skeletal-concentration layers of disarticulated trilobite carapaces do not show any [Palaeontology, Vol. 40, Part 2, 1997, pp. 385-396] © The Palaeontological Association 386 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Location of the fossil site of Rotadiscus guizhouensis Zhao and Zhu, 1994 above the village Balang near Taijiang in Guizhou province of China, a, rock column of the most fossiliferous part of bed 24 of the Kaili Formation at this site, B, section along the exposure with the most fossiliferous site arrowed, c, geological map of the area showing distribution of the Kaili Formation and adjacent lithological units, position of the section b indicated. D, map showing location of Yangtze Platform. features of transportation, preferential orientation or sorting (Zhu and Zhao 1995). There are only a few bedding surfaces on which the trace fossil Treptichnus and lingulide brachiopods occur. Some Rotadiscus-covered bedding planes occur about 6 m below the first carbonate beds in the uppermost part of the formation (Zhu and Zhao 1995). Two horizons with mass occurrences of the Rotadiscus discs are clearly sedimentary discontinuity surfaces. Similar surfaces are also known in this part of the section with numerous articulated DZIK ET A L. \ CAMBRIAN ELDONIOID 387 eocrinoids (Gogia). In the latter case, the orientation of the echinoderm tests shows that they were killed instantaneously by a turbiditic flow of sediment, with a basal layer, a few millimetres thick, of coarser mud showing grain gradation (Zhu and Zhao 1995). The fate of the Rotadiscus associations was similar, although they are usually covered by a finer sediment. The silty mudstone beds deposited above other such surfaces, usually less than 20 mm thick, may show unidirectional current marks at the basal surface that indicate rapid sedimentation (Zhu and Zhao 1995, p. 8). The original orientation of well-preserved discs scattered over the sedimentary discontinuity surfaces within bed 24, whenever observed in the field, was convex side down. Yuan et al. (1995) correlated the uppermost part of the Kaili Formation, containing the pelagic eodiscid Peronopsis majiangensis , with the North American Plagiura-Kochaspis Zone, that is with the lowest Middle Cambrian. The Kaili Formation is thus significantly older than the Burgess Shale, being closer in age to the Polish occurrences of eldonioids, in the first Baltic trilobite zone (Eccaparadoxides insularis ) of the Middle Cambrian (Masiak and Zylinska 1994). ANATOMY OF ROTADISCUS The Kaili specimens of Rotadiscus guizhouensis Zhao and Zhu, 1994 commonly show presumed carbonized remains of soft organs, mostly a U-shaped gut. The specimens are flattened and no internal filling with sediment has been observed, with the possible exception of the specimen ZPAL CH.E/5 where there seems to be a thin layer of mudstone between the level with the imprint of the intestine and the disc wall (Text-fig. 2f). Usually the darkest area, corresponding presumably to the muscular wall of the intestine and its organic-rich content, is followed along both sides by a paler zone delimited externally by dark lines (Text-figs 2b, f, 3a). This corresponds to the structure called the coiled sac by Chen et al. (19956), which seems to be at least analogous to the peritoneum. The most important difference in the internal organizations of R. guizhouensis in respect to the Chengjiang and Burgess Shale material is that the intestine was significantly shorter in the Kaili species, with oral and anal openings relatively distant to each other, so that the intestine is truly U- shaped. The posterior end of the intestinal structure terminates abruptly in the Kaili material, which seems to indicate that the posterior part of the digestive tract was empty and not well enough muscularized to impart carbon staining. The dark stain in the anterior part of the intestine is increasingly paler towards the mouth, but some of its structures can be traced to where the two sets of tentacles are located at the corners of the mouth opening (Text-fig. 2d). The preservation of the fossil is not good enough to show details of the lophophore organization, apart from possibly two bunches of tentacles emerging directly from each of the two bases. Some traces of radial canals (or perhaps muscles) are also visible in the Kaili material (Text-fig. 2f). In two specimens a central cavity is recorded by a pale staining. The body of Rotadiscus was disc-shaped but not completely flat. Its minimum thickness can be inferred from the diameter of the intestine, which was by no means thin and, together with the peritoneum, corresponds to about one-third of the disc radius (Text-fig. 2e-f). Its dorsal surface was apparently smooth, as no interference of two sets of concentric growth lines systems is observed, as would be the case if the dorsal surface was also sclerotized. The ontogeny of Rotadiscus guizhouensis was more complex than in the ancestral R. grandis. The central area of the disc, a couple of millimetres in diameter, seems to be smooth when well preserved (Text-fig. 3c). In numerous cases it bears one or two prominent marginal growth wrinkles (Text- fig. 2a). At this stage, the animal was about 3 mm in diameter. External to these wrinkles, sharp radial furrows develop on the external surface, about 40 in number. They allow easy recognition of the disc side on specimens that are not completely flattened. These furrows make R. guizhouensis similar to the coeval Polish Velwnbrella czarnockii Stasinska, 1960 at this stage of ontogeny, but are almost twice as numerous. The furrows disappear gradually at disc diameters ranging from 20 to 40 mm. The subsequent stage in ontogeny is characterized by a lack of any radial ornament, but the concentric step-like growth lines become more and more prominent and numerous, indicating 388 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Rotadiscus guizhouensis Zhao and Zhu, 1994; lower Middle Cambrian, Kaili Formation, Balang; specimens with preserved intestine, photographed whitened with ammonium chloride sublimate (a, c, e) and submerged under alcohol (b, d, f). The oval shape of this and following specimens is an effect of tectonic deformation, a-d, inner and outer side of the disc ZPAL CH.E/3b. a (counterpart and part, respectively), e-f, outer side of the disc ZPAL CH.E/5. All x L5. maturity. The growth of the Rotadiscus discs was thus finite, the specimens becoming adult usually at a diameter of about 40 mm, although a specimen with a diameter of 91 mm has been found in bed 24 (the second largest in size, from bed 21, is 69 mm). Only these extremely large specimens DZIK ET AL. : CAMBRIAN ELDONIOID 389 text-fig. 3. Rotadiscus guizhouensis Zhao and Zhu, 1994; lower Middle Cambrian, Kaili Formation, Balang; specimens with epizoans ‘ Chuandianella subovata Yuan and Huang, 1994. a-b, outer side of the disc ZPAL CH.E/19 with preserved intestine, photographed submerged under alcohol (a) and whitened with ammonium chloride sublimate (b); note isolated large epizoans and repair of the disc in its lower part, c, part of slab ZPAL CH.E/1 with the sedimentary discontinuity surface covered with discs, which may overlap together with their numerous epizoans; note scattered complete and disarticulated carapaces of the eodiscid Pagetia. All x F5. approach the mean sizes of the Chengjiang Eldonia and Rotadiscus , or the Polish Velumbrella. The giant ‘ Brzechovia' , with its 180 mm diameter discs (Dzik 1991), is significantly larger. AFFINITIES OF ROTADISCUS The first described species of the group was Eldonia ludwigi Walcott, 191 1 from the Mid Cambrian Burgess Shale, originally interpreted by Walcott (1911) as a holothurian, a view supported by A. H. Clark (1912, 1913), H. L. Clark (1912) and Durham (1974). According to Madsen (1956, 1957, 1962), Eldonia was a siphonophore, whilst Lemche (1960) considered it to be a coelenterate medusa. The specimens of Eldonia ludwigi , and even better preserved material of the related E. 390 PALAEONTOLOGY, VOLUME 40 eumorpha (Sun and Hou, 1987) from the Early Cambrian Chengjiang site, provided anatomical evidence for a more detailed zoological classification of the eldonioids (Dzik 1991; Chen et al. 1995 6). Another well-known Chengjiang species is Rotadiscus grcmdis Sun and Hou, 1987. There are no signs of any pentameral or tetrameral symmetry in the anatomy of Eldonia, nor is there any reason to suppose the presence of a mesodermal calcitic skeleton; so both the holothurian and scyphozoan affinities of this animal are difficult to support. A U-shaped intestine (Text-figs 2-3), lophophore, and an ectodermal, marginally accreted skeleton are features of the lophophorates. Such an attribution was proposed for the eldonioids (in the rank of class) by Dzik (1991). The anatomical data on R. guizhouensis are consistent with those on the Burgess Shale and Chengjiang species, which shows that the body plan of these organisms was rather uniform. The relatively simple organization of this species, with a strictly U-shaped digestive tract and the disc inferred to be more conical than in other species (Text-fig. 5), may reflect its primitiveness with respect to the larger and more complex eldonioids. LIMITS OF MOBILITY IN ROTADISCUS The eldonioids were widespread in Early and Mid Cambrian seas and are known from strata that originated in various environments, ranging from the open marine clays of the Burgess Shale in British Columbia (Walcott 1911) to the coarse sands of the Ociesgki Formation in Poland (Stasinska 1960; Dzik 1991 ; Masiak and Zylinska 1994). Different morphologies are represented in each of these localities; but all three-dimensionally preserved eldonioids show growth lines at least on the convex surface of their discs. In Eldonia (Chen et al. 1995a, 19956) the disc was probably only lightly sclerotized, but cuticle accretion was clearly marginal at both ventral and dorsal sides of the body. The accretional mode of growth implies a certain stiffness of the cuticle and, even if it was not mineralized and thereby flexible in Eldonia , its elasticity was definitely low. It is thus unlikely that the circumference of the disc could be reduced significantly by muscle contraction that would enable a medusa-like locomotion. The U-shaped intestine, enveloped in a kind of peritoneum (Chen et al. 19956; see also Text-fig. 3a), implies a coelomate anatomical organization and makes jet propulsion by ejecting water from the gastric cavity equally unlikely. No anatomical structure that could serve as a hydrostatic organ was identified in the excellently preserved Chengjiang specimens. This would be necessary to make the animal buoyant and stable in water with its heavy skeletonized part, lacking any gas-filled chambers, upwards (against the inferred position of the centre of gravity). Thus, a medusoid mode of life cannot be directly inferred from, and seems to be contradicted by, the anatomical organization of Eldonia. There is no particular reason to believe that Eldonia was different in this respect from other genera of the Eldonioidea that have basically the same body form, but differ mostly in the degree of sclerotization (and preservation) of their body cover. The stiff skeleton of Velumbrella from Poland fragmented into angular pieces, without any plastic deformation, in a coarse sandstone matrix (Dzik 1991, fig. 3b; Masiak and Zylinska 1994); the Chengjiang Rotadiscus ‘ formed cracks on compaction when compared with wrinkling and folding observed in Eldonia ’ (Chen et al. 1995n). In the Kaili species of Rotadiscus , repair of the disc skeleton can be observed in rare cases (Text-fig. 3b) - apparently a result of unsuccessful predation of the disc margin. The subsequent regeneration proceeded in the same way as in other marginally secreted shells of molluscs or brachiopods. In Recent pelagic faunas no organism of comparable organization, with strongly sclerotized, marginally accreted skeleton covering only one side of the body, is known. Such skeletal features of the eldonioids are thus hardly compatible with their allegedly ‘medusoid’ mode of life. The pattern of epibiontic overgrowth of the discs, discussed below, makes this even less likely. EPIZOANS Rotadiscus is the most heavily sclerotized eldonioid, although its cuticle was probably not mineralized. In the Kaili Formation, preservation of the Rotadiscus disc is different from that of DZIK ET AL.\ CAMBRIAN ELDON IOI D 391 DISTANCE FROM DISC CENTRE DISTANCE FROM DISC MARGIN text-fig. 4. Scattergrams of the epizoans shell length against distance from the centre (a) or margin (b) of the disc of Rotadiscus guizhouensis Zhao and Zhu, 1994 from the lower Middle Cambrian Kaili Formation at Balang (based on 22 well-preserved, epibiont-bearing discs from the GTB collection; 36 other discs of comparable preservation from the same collection lack epibionts); incomplete preservation of weakly sclerotized disc margin in some specimens may have resulted in some distortion of the pattern presented by the scattergram b, but note that the shells cluster along the disc margin irrespective of its size. associated originally calcific trilobite carapaces and aragonitic mollusc conchs. The discs are always more strongly compressed but, although no breakage or fragmentation has been observed, folds on the disc’s surface are always sharply angular and linear. This indicates some rigidity of the skeleton, which was thus, at least potentially, a suitable place for the settlement of epibionts. In fact, filamentous algal thalli radiating from the disc margins in some specimens are suggestive of attachment to its skeletonized surface. Along with algae, which are usually difficult to discern, more prominent objects do occur in association with the Rotadiscus discs in the Kaili Formation. Scattered over the discontinuity surfaces in the upper part of the section, the marginal parts of discs are covered by small oval shells. The zoological affinity of these shells remains obscure. They were originally described as the bradoriid ostracode Chuandianellal subovata Yuan and Huang, 1994 and C.? linguiformis Yuan and Huang, 1994, although no specimen with two valves in articulation has been found in the Kaili Formation. Virtually all the shells are crushed due to sediment compaction, resulting in the original thickness and outline of the shells being deformed. Only in layers of the mudstone with a slightly higher content of calcium carbonate do they preserve a shape and convexity close to the original (Text-fig. 3b). In such cases the shell outline is ovoid; the shells are bilaterally symmetrical. The rugation, commonly visible on the shell surface, probably represents marginal growth increments. Their arrangement suggests that either at earlier growth stages the shell was more circular and its growth attenuated towards the narrower end of the shell or, more probably, that the shells originally were very convex and their proximal parts are compressed obliquely, resulting in shape distortion. The latter is consistent with small and large epizoans having the same shell outline. In two specimens (ZPAL CH.E/2 and 13) the epizoans located close to the centre of the disc have their interiors filled with siliceous and clay minerals. The original convexity of the shell is partially preserved. The base remains completely flat, its surface following the rugation of the disc. The 392 PALAEONTOLOGY, VOLUME 40 DZIK ET AL. \ CAMBRIAN ELDONIOID 393 margin of the shell was apparently rather thick - a feature shown also by the empty cavities left after shell margins dissolved in some other specimens. It is thus apparent that the epizoan shells were originally not organic. When compared with Rotadiscus discs and associated trilobites, the resistance of the epizoan shells against compaction was intermediate between these two kinds of fossils. The epizoan shells evidently were not calcitic because agnostid carapaces commonly preserve more or less altered original calcareous walls. Perhaps aragonite was involved in the shell construction. Probably in the two better preserved specimens mentioned above, the organic matter of the host Rotadiscus protected the aragonite during early diagenesis and only later, after compaction, was it transformed into calcite and replaced by other minerals. Rarely, some specimens of the epizoan occur disassociated from the Rotadiscus discs (for instance, the largest shells in specimen ZPAL CH.E/19; Text-hg 3b). Occasionally such isolated specimens bear prominent ridges that run parallel to each other, without any correspondence to the inferred mode of growth of the shell (see Yuan and Huang 1994, pi. 1, fig. 8). Apparently this is a xenomorphic ornament reproducing rugation of the Rotadiscus disc margin. This, together with the already noted flatness and rugation of the basal surface of the epizoans, indicates that the secreting margin of the epizoan shell strictly followed the substrate during growth. Such could be a result of either cementation to the disc surface or at least very strong attachment with a pedicle or byssus. The shells of the Rotadiscus epizoans are always univalved. This refers to those that had apparently fallen from the disc surface, prior to or during the catastrophic covering of the bottom by sediment, but also to specimens still attached to the disc. If they were cemented to the disc, the basal valve must have been very thin and probably inconspicuous. The bilateral symmetry of the shells, their probable marginal accretion and possibly cementing mode of life, make arthropod affinities of the epizoans seem unlikely. Their shape and inferred mineralogical composition is similar to that of the Early Cambrian inarticulate obolellid brachiopods (see Goryansky and Popov 1985; Geyer 1994) and the monoplacophorans. Similar morphologies and a probable calcareous shell mineralogy also characterize the problematical Cambrian Apistoconcha , an organism of possible brachiopod affinities (Bengtson et al. 1990). No cementing Cambrian brachiopods are known, however. The only known Mid Cambrian bivalved organism that cemented to the substrate was the operculate coral Cothonion (Jell and Jell 1976). Unfortunately, our knowledge of the morphology of the Kaili epizoans, as well as their mode of attachment to Rotadiscus discs, frustrates further discussion of their affinities. PATTERN OF OVERGROWTH Attachment of the epizoan shells to the Rotadiscus discs was rather firm. Discs that happen to overlap always bear their own, clearly identifiable, aureole of marginal shells (Text-fig. 3c). Isolated dispersed shells are rare (a few per cent, of the total number) and their quite occasional occurrence cannot result from their being removed from the central part of the Rotadiscus discs. In all the numerous specimens that are not so compressed as to obliterate the relationship of the disc surface to associated objects, they cover the convex surface of the disc. Apparently, the fiat surface of Rotadiscus was not stiff enough to allow attachment. At least seven epibiont-bearing discs in the studied collection have their internal soft organs preserved (Text-fig. 3a) which proves that Rotadiscus was overgrown while alive. The most puzzling feature of the epibiontic cover of Rotadiscus discs is the size distribution of the shells in relation to the disc margin. If the entire surface of Rotadiscus was exposed to sea water and the epizoan larvae settled on it randomly, then the largest specimens would be expected close to the text-fig. 5. Rotadiscus guizhouensis Zhao and Zhu, 1994; lower Middle Cambrian Kaili Formation; restoration of the disc apical surface with attached epizoans ' Chuandianella' subovata Yuan and Huang, 1994 (a) and a disc in its proposed life position sectioned to show the location of the intestine and central cavity (b). 394 PALAEONTOLOGY, VOLUME 40 disc centre, that is on its oldest part. In reality, the size-frequency distribution in respect to the disc centre is the opposite; their size being positively correlated with the distance from the disc centre (Text-fig. 4a). Thus, the largest specimens are invariably located at the disc margin. This means that larger discs have generally larger epizoans at their margins, and size-frequency distribution with respect to the distance from the disc margin does not show any apparent pattern (Text-fig. 4b). Small epizoans may thus occur in association with larger ones, but usually in more central positions. Apparently, the epizoans’ larvae settled only in proximity to the disc margin where they are concentrated irrespective of the ontogenetic age of the disc (expressed by its diameter; Text-fig 4b). The terminal growth of Rotadiscus (evidenced by the condensation of growth lines) suggests that marginal areas of the disc provided longer time spans for the epizoans' settlement than the central areas. Thus, the epizoans crowd close to the margin of mature discs, as if colonization was seasonal, which was probably not the case. Even if the epizoans preferentially settled at the disc margin, their size-frequency distribution cannot be otherwise easily explained if the epibiontic nature of the shells is accepted. The anatomy of the eldonioids, with their discs covered by a cuticle (thus unable to produce ciliary feeding streams of water) and a centrally located smaller tentacular apparatus (Chen et al. 1995A), does not make the disc margin more suitable than the disc centre for life as a filter feeder, assuming that the disc surface was completely exposed to the sea water. The only reasonable explanation for this pattern is that it resulted from a peculiar pattern of epizoan mortality. Rare epizoan larvae that settled on the marginal parts of discs at an earlier stage were subsequently killed by the anoxic conditions under the disc as it lay passively on the black mud. They are relatively uncommon (Text-fig. 4), probably because of the relatively fast growth rate of young discs. The disc margin was a suitable place for epibiont settlement only after its accretion rate decreased and finally ceased with the animal’s maturation. CONCLUSIONS We propose that the main factor controlling the distribution of the ‘ Chuandianella ’ subovata epizoans was the anoxic microenvironment underneath the Rotadiscus guizhouensis discs that lay passively on the organic-rich mud. The bottom environment was definitely anoxic during the deposition of the Kaili Formation, as the strata contain abundant articulated pelagic and benthic trilobites and other excellently preserved fossils, with originally non-mineralized skeletons. The rare Rotadiscus discs occurring within the parts of the sequence, where a higher sedimentation rate presumably resulted in disc margins not being exposed by winnowing, are commonly devoid of any epibionts. Such is, in fact, the case in the largest known disc (91 mm in diameter). There are horizons, however, where all the discs bear epibionts. These are bedding planes corresponding to sedimentary breaks. On some of them shoals of Rotadiscus developed; other surfaces are covered by numerous gogiid echinoderms. They flourished, with eldonioid disc margins free of sediment cover and encrusted by various epibionts (including filamentous algae, observed in several specimens), until being instantaneously killed by a turbiditic flow of sediment. The type of epibiont growth associated with episodes of lowered sedimentation rate makes it unlikely that the shells (apparently heavy) encrusted the discs as pelagic forms in the water column. The reasoning presented above remains valid even if one assumes the less likely possibility (contradicted by the xenomorphic ornament and rugated bases in some of the epizoans) that the epizoans were able to change their position on the disc and migrated with the growth of the host disc. Acknowledgements. The present paper is based on collections assembled during a field expedition to the Kaili locality financed by the Polish Committee of Scientific Research (grant No. 6 P04D 046 08) and organized by teams from the Giuzhou Institute of Technology and Instytut Palaeobiologii PAN. We are grateful to Simon Conway Morris (University of Cambridge), Adolf Seilacher (Yale University) and an anonymous reviewer, who provided numerous helpful suggestions and kindly improved the style of this paper. DZIK ET AL.: CAMBRIAN ELDONIOID 395 REFERENCES bengtson, s., conway morris, s., cooper, b. j., jell, p. A. and runnegar, b. n. 1990. Early Cambrian fossils from South Australia. Memoir of the Association of Australasian Palaeontologists , 9, 1-394. chen jun-yuan, zhu mao-yan and zhou gui-qing 1995a. Anatomy and phylogenetic relationships of the medusiform Eldonioidea. 9-10. In chen jun-yuan, edgecombe, b. and ramskold, l. (eds). International Cambrian Explosion Symposium ( April , 1995, Nanjing) ( Programme and Abstracts). Nanjing Institute of Geology and Palaeontology. Academia Sinica, 48 pp. — 19956. The Early Cambrian Tnedusoid’ Eldonia from the Chengjiang fauna of China. Acta Palaeontologica Polonica, 40, 213-244. Clark, A. h. 1912. Restoration of the genus Eldonia , a genus of free swimming holothurians from the Middle Cambrian. Zoologisches Anzeiger, 39, 723-725. 1913. Cambrian holothurian. American Naturalist, 47, 488-507. Clark, h. L. 1912. Fossil holothurian. Science, 35, 274—278. Durham, w. j. 1974. Systematic position of Eldonia ludwigi Walcott. Journal of Paleontology, 48, 750-755. dzik, J. 1991. Is fossil evidence consistent with traditional views of the early metazoan phylogeny? 47-56. In simonetta, a. m. and conway morris, s. (eds). The early evolution of Metazoa and the significance of problematic taxa. Cambridge University Press, Cambridge, 296 pp. geyer, G. 1994. A new obolellid brachiopod from the Lower Cambrian of Morocco. Journal of Paleontology, 68, 995-1002. Goryansky, v. Y. and popov, L. E. 1985. Morfologia, sistematiteskoje polozeme i proischozdienie bezzam- kovych brachiopod s karbonatnoj rakovinoj. Paleontologicheskij Zhurnal, 1985 (3), 3-14. [In Russian]. jell, p. a. and jell, J. s. 1976. Early Middle Cambrian corals from western New South Wales. Alcheringa , 1, 181-195. madsen, F. j. 1956. Eldonia, a Cambrian siphonophore formerly interpreted as a holothurian (sic). Videnskabelige Meddelelser fra Dansk naturhistorisk Forening i Kobenhavn , 118, 7-14. — 1957. On Walcott’s supposed Cambrian holothurians. Journal of Paleontology, 31, 281-282. — 1962. The systematic position of the Middle Cambrian fossil Eldonia. Meddelelser fra Dansk geologisk Forening, 15, 87-89. masiak, m. and zylinska, a. 1994. Burgess Shale-type fossils in Cambrian sandstones of the Holy Cross Mountains. Acta Palaeontologica Polonica, 39, 329-340. otto, m. 1994. Zur Frage der ‘ Weichteierhaltung’ lm Hunsriickschiefer. Geologica et Palaeontologica, 28, 45-63. rigby, s. and milsom, c. 1996. Benthic origins of zooplankton: an environmentally determined macro- evolutionary effect. Geology, 24, 52-54. seilacher, a. 1989. Vendozoa: organismic constructions in the Proterozoic biosphere. Lethaia , 22, 229-239. Stanley, G. d. jr 1986. Chondrophorine hydrozoans as problematic fossils. 68-86. In hoffman, a. and nitecki, m. h. (eds). Problematic fossil taxa. Oxford University Press, New York, 267 pp. — and sturmer, w. 1983. The first fossil ctenophore from the Lower Devonian of West Germany. Nature, 303, 518-520. 1987. A new fossil ctenophore discovered by X-rays. Nature, 328, 61-62. stasinska, a. 1960. Velumbrella czarnockii n. gen. n. sp. Meduse du Cambrien inferieur des Monts de Sainte- Croix. Acta Palaeontologica Polonica, 5, 337—396. sun wei-guo and hou xian-guang 1987. Early Cambrian medusae from Changjiang, Yunnan, China. Acta Palaeontologica Sinica, 26, 257-271. walcott, c. d. 1911. Cambrian geology and paleontology 11, no. 3 Middle Cambrian holothurian and medusae. Smithsonian Miscellaneous Collections, 57, 41-68. yuang jinliang and huang you-zhuang 1994. Preliminary report on non-trilobite arthropods from Lower-Middle Cambrian Kaili Formation of southeastern Guizhou, South China. Acta Palaeontologica Sinica. 33. 329-334. — zhao yuanlong and wang youzhuang 1995. Trilobite zonation and correlation of the Kaili Formation. 9-10. In chen jun-yuan, edgecombe, G. and ramskold, l. (eds). International Cambrian Explosion Symposium (April, 1995, Nanjing) (Programme and Abstracts). Nanjing Institute of Geology and Palaeontology, Academia Sinica, 48 pp. zhao yuan-long and zhu mao-yan 1994. Medusiform fossils of Kaili fauna from Taijiang, Guizhou. Acta Palaeontologica Sinica, 33, 272-280. 396 PALAEONTOLOGY, VOLUME 40 zhu mao-yan and zhao yuan-long 1995. Geology of the Kaili fossil biota. 42. In zhang junming, zhu maoyan, Li guoxiang, ZHOU chuanming and Yu ziye (eds). Excursion Guide for Internal Symposium , Early Cambrian Breakthrough , Environment , and Mineral Resources. Nanjing Institute of Geology and Palaeontology, Academia Sinica. JERZY DZIK Instytut Paleobiologii PAN Twarda 51, 00-114 Warszawa, Poland ZHAO YUANLONG Department of Geology Guizhou Institute of Technology Guiyang 660003, Guizhou, China ZHU MAOYAN Nanjing Institute of Geology and Palaeontology Typescript received 21 May 1996 Academia Sinica Revised typescript received 24 October 1996 Chi-ming-ssu, Nanjing 210008, China. LATE ORDOVICIAN TRI LOBITES FROM SOUTHERN THAILAND by RICHARD A. FORTEY Abstract. A rich and well-preserved trilobite fauna is described from the upper Ordovician (Caradoc) Pa Kae Formation, Satun Province, southern Thailand. This is the first diverse trilobite fauna of this age to be described from the Shan Thai (Sibumasu) terrane. The fauna represents an outer shelf assemblage of 39 species, dominated by Ovalocephalus , nileids and remopleuridids. It is identical even at the species level to faunas described from the Pagoda Limestone of southern China, indicating that the Shan Thai terrane cannot have been far removed geographically from this part of China in the late Ordovician. The distribution of the Ovalocephalus fauna proves that there are also wider faunal links with western Gondwana and Scandinavia. Most of the species have been previously named from China, but new information is presented for several of them. Examination of the agnostid Arthrorhachis latelimbata proves that cell polygons are a much hner-scale structure than reticulate sculpture. The telephinid Telephina convexa has holochroal eyes constructed of perfectly square lenses: the possibility that these represent reflection superposition eyes is considered. Two new species are proposed : Sculptaspis pulcherrima and Ovalocephalus plewesae. Very little has been published on Ordovician trilobites from Thailand (Stait el al. 1984), and no late Ordovician fauna has been described hitherto. The Pa Kae Formation (Wongwanich et al. 1990) is a distinctive unit of red-weathering limestones cropping out in the southernmost Satun Province of Thailand, close to the Malaysian border (Text-fig. 1). Dr C. Burrett and T. Wongwanich discovered trilobites in this unit in 1986, and in 1987 the author visited the type sec- tion (Text-fig. 2), which was then entirely clear of vegetation, and collected many more specimens. A second visit was made in 1994, at which time the lower part of the section had already become obscured under a vigorous growth of rubber trees. Not only did the fauna prove to be a new one for the Shan Thai terrane, but it is also well preserved, which makes a systematic account worthwhile. Previous descriptions of the species have mostly been in Chinese journals which are hard to obtain in the West. Nearly all the trilobite genera are described from the Shan Thai block for the first time, and several are recorded for the first time outside the Yangtze Platform. The palaeogeographical position of the Shan Thai terrane in the Palaeozoic is the subject of controversy (Metcalfe 1992), and the palaeobiogeography of the trilobite faunas contributes to our understanding of the position of this region in the late Ordovician. THE PA KAE FORMATION The Pa Kae Formation crops out in a small area of Satun Province, southern Thailand (Wongwanich et al. 1990). The type section (Text-fig. 1) is a small hill about half a kilometre east of the main road (Route 4078) along the track to the village of Ban Pa Kae, which is itself about 10 km north of the town of Langu, La Ngu District. On the 1 : 50,000 map sheet, Amphoe Langu (reference 49221), the outcrop occupies the area between grid references 8583 7122 and 8592 7085. This is at 6° 58' 25" N, 99° 46' 42" E. The Pa Kae Formation is the uppermost limestone formation in the Ordovician sequence of southern Thailand. Almost all the collections described here were made from the type section, which can be identified by conspicuous red bluffs by the side of the track. IPalaeontology, Vol. 40, Part 2, 1997, pp. 397^449, 10 pis) © The Palaeontological Association PALAEONTOLOGY, VOLUME 40 398 text-fig. I . Locality map of the Pa Kae Formation in southern Thailand (right) with South-east Asian terranes indicated on left hand diagram, with modern national boundaries. Field characteristics. The lower part of the Pa Kae Formation comprises massive pelmatozoan limestones from which no trilobites have been collected. These produce a prominent topographical feature, and the base of the section shown in Text-figure 2 is taken above these conspicuous beds. The overlying strata are comparatively well-bedded, red-weathering muddy limestones which form a series of small bluffs: bed-by-bed collecting was possible through much of this section. The limestones are notable for their syneresis cracks which form conspicuous polygons standing proud from the surrounding limestone. Some particularly ferruginous horizons contain patches of haematite, a mineral which also commonly lines stylolite seams. Although the impure limestones often appear dense and structureless, where there is appropriate weathering there are indications of extensive bioturbation. Orthoconic nautiloids occur in several beds. Instead of being concentrated into lenses or particular beds, the trilobites appear to be scattered sparingly throughout the limestones. Much breaking of rock is required to obtain a good number of specimens. In most beds they are exquisitely preserved (see, for example. Text-fig. 4). Although fully articulated specimens have not been collected, there is a fair number of cephala which retain their free cheeks; this is an indication that the trilobites have probably not been greatly reworked, and that the species found probably lived together in situ. The comparatively large number of specimens about 10 mm long or less in the collections may be partly an artefact of the collecting technique. The rocks are so hard and even grained that the trilobites do not crack out. The collections were made by progressively breaking collected limestone blocks to cubes of side 10 mm and then manually preparing any specimens showing on the surface. This method provides a good sample per volume, but also biases against larger specimens. There is a sparse fauna of brachiopods accompanying the trilobites which is referable to the Foliomena fauna (Cocks and Rong 1988, p. 67). Depositional environment . The lithology and fauna (see below) of the Pa Kae Formation appear to be identical to that of the Pagoda Limestone Formation, which is widespread across central and south-western China. In the past, this formation has been regarded as of shallow water origin, because the polygonal cracks abundantly developed within it have been thought to be desiccation FORTEY: ORDOVICIAN TRILOBITES 399 cracks. Ji (1985) has pointed out that these are better interpreted as syneresis cracks. These are formed in clay-rich sediments beneath continuous water cover (Burst 1965). The Foliomena brachiopod fauna is regarded as the deepest water one in the latest Ordovician (Cocks and Rong 1988), and Boucot (pers. comm. 1994) places it in his Benthic Assemblage 4. Ji (1985) states that the water depth under which the Pagoda Limestone was deposited may have been 70-100 m. The trilobites represent a rich fauna: Ovalocephalus is the commonest genus, ParaphillipsineUa , Remopleurella , agnostids and nileids are numerous, while the other genera are represented by small numbers of specimens. However, cyclopygids and Telephina are present, both elements of an open ocean pelagic fauna (Fortey 1985). Price and Magor (1984) portrayed a shallow to deep water biofacies profile for the Ashgill of North Wales: an assemblage dominated by Dindymene , cyclopygids and telephinids typified a deep water, but not deepest, biofacies (where a dominance of cyclopygids and blind trilobites was typical). Nileids and raphiophorids occupied a similar depth zone to that of the Pa Kae Formation in the earlier Ordovician (Fortey 1975a). Atheloptic trilobites having reduced eyes (Fortey and Owens 1987) are uncommon in the Pa Kae Formation, and the benthic part of the fauna is dominated by trilobites with normal eyes. It seems very likely, therefore, that the sea floor was within the photic zone, and probably at a depth of 200 m or less. This seems to agree with Ji's (1985) estimate, based on sedimentary features of the Pagoda Limestone Formation. Benthic brachiopods are present, along with the trilobites, and there are also gastropods, bivalves and rare echinoderms. This rich fauna, combined with the evidence for bioturbation (possibly also the oxidation state of the iron compounds, if original) surely indicates a well oxygenated sea floor. This may be the reason why graptolites are not preserved, although they are numerous in the black shales succeeding the Pa Kae Formation (Wongwanich et al. 1990). AGE AND CORRELATION OF THE PA KAE FORMATION The fauna of the Pa Kae Formation may be divided into a lower one and an upper one (Text-fig. 2) ; the latter yields more prolific fossils. However, several species, including Sphaerexochus fibrisulcatus and Ovalocephalus ovatus , range throughout, and there are overlapping species ranges in the mid-part of the collected section, both of which suggest continuous deposition over a comparatively short period. With rare species, one must be cautious about sampling effects artificially truncating the true length of ranges (Strauss and Sadler 1989). However, the Pa Kae Formation is apparently confacial throughout, and there are stratigraphically related species changes through the section in the agnostids, and in certain genera (e.g. Ovalocephalus , Pauderia , Remopleurella ), which suggest that the changes observed are of biostratigraphical value. The lower fauna shares several species with the small fauna described by Kobayashi and Hamada (1978) from Langkawi Island, not far to the south. The species Geragnostus perconvexus and Lonchodomas rhombeus are significant in proving this correlation, while Nileus species are also similar. Neither of the first two species has been recorded from the Pagoda Limestone Formation in southern China. Kobayashi and Hamada (1978, p. 1) were imprecise as to the age of the Langkawi Island fauna stating that it was ‘in a range from middle Caradoc to early Ashgill’. There are two species definitely found both in the lower fauna and in Bed 12 of the Chedao Formation of Gansu Province, China (Zhou and Dean 1986): Ovalocephalus p/ewesae sp. nov. (identified by Zhou and Dean as O. kelteri Koroleva) and ParaphillipsineUa globosa. In the systematic section, I also suggest that Lonchodomas nanus Zhou and Nileus huanxianensis Zhou from the Chedao Formation may prove to be junior synonyms of species from the Pa Kae Formation. Zhou and Dean (1986, p. 744) listed many additional species of trilobites and associated cephalopods from Bed 12 as evidence for its correlation with the Pagoda Limestone Formation of the Yangtze region, which is regarded as of Caradoc age (Zheng et al. 1983; Chen et al. 1995). Hence a correlation of the lower fauna with some part of the Pagoda Limestone Formation is likely. The Langkawi Island faunule is therefore also likely to be Caradoc. £0 _) ‘-O % \ no : ' no T3 > -s — NO f -j&9 no ,->J r? u . = ?~ § sr s-s-; ^o'-SffSi S3 NT3 Cl — 3 < -&aP „ O _ £ 3 LX — 2 3* __ S n> -0* o r* g- " — •“ “ 0-0 2 Er =f (TO 03 O CL o' <. ; 3? Cl • S’rS* O ^ ?Js. ’ §■ ?f -0 r? ^ >3 : 3 °° g I ! 2 3 <.■§ I § n § c i n> a.05 x ! 1° fl §=! ? 3 ~ ~ » i bS?o? p g ill ! c=§^5-; o ■§ S' § ~r. 5 ^ r” L1?? § lf< a o s ip, ^ o >3 r. F'b m 1 o ■> 03 oq T3 C ■ > 1: i- §■ r.-s 3 ^ U 3 ► S O 03 o ;S?3 i S'a.g-a §; O-g • re 5: O agre |o " 5lr§ | 9 o S o " 7 S » 5' «-► 3 *0 ’'OfTO 5* 3 ^ " 1 1 1. o 3 i f Qp3^35pg 5 " O rC ^ O ^ ' O o ? 5- o 3 o 7* Cro a. os 3 < 03 a 8* O O* « 5-«» jj. 03 ~ 1 S’ S' - 2 : S g p j _ 0. rj 3 O -• d- 2 3 S o C ? o n '!i ;• c/3 ! £L 00 ‘ - 3. ac - S » g n o n 3 :r “' I O § 5^ I 7 |M 5 |B'i. :• =r n, 3 j fro 3 O 3 n ^ c _ . ^ k re * 2 3 ^-TD >? oj c n g | ?3 "o 2 r~ 3 3 ttq c*j 2 o ~ s s S <.3 ® 2. r^. ” ^ ^ " < C: % • =r o ^ a 00 o, -3-" P S’ O' t O _ ^ 5‘ S 3 J 5 u _. ft o S Ci 03 3 -> -t or =: no _ >1 3?8.§ 3- S’ P W | CTO Cl p 3 S ? S 0 s o L. 3 2: re o_to p1 5" g-m I ^ gl | ?■§ 5; S ll' = 2-§.| 2 » 5 S 3 E« 3 ^ w 3 z. =: 3 o 2 > s' 3 3 S § § g" ! o x H o ; 5. S |. § re & E S' a*. 2 *a» n £. j I & g | g* 8 | *-T* Cfl' ^ O “ • ff 03 H; 1 ~ ^ o..§ 5’ ; J'sl-'S s- N 03 o o Ci VO P1 'Si' re g , I s-cro u. a g 2 j? *5- c &• r rro a 3 1 text-fig. 2. Section through the upper part of the Pa Kae Formation at its type locality showing ranges of trilobites within the section. Base of the measured section is taken above massive prinoidal limestones forming the lower part of the formation. 3 o o o o Qvalocephalus ovatus Sphaerexochus fibrisulcatus Nile us transversus Elongatanileus convexus • Arthrorhachis latelimbata • Arthrorhachis sinensis Telephina convexa • Rorringtonia cf. lepida • Nileus malayensis • • Uchas sp. • Microparia cf. speciosa Hadromerus xiushanensis Lonchodomas rhombeus • Panderia orbiculata Panderia migratoria Cekovia striata Cekovia transversa Sculptaspis pulcherrima Paraphillipsinella globosa_ Qedicybele sp. nov. A Qvalocephalus plewesae Parisoceraurus rectangularis Geragnostus perconvexus Hanjiangaspis fibrisulcata _ Paraphillipsinella nanjiangensis Oedicybele sulcata • Cyclopyge recurva Parvigena plana • Corrugatagnostus jiangshanensis • Taklamakania sp. • Quyuania cf. ziguiensis • Cyamella sp. 1 • • Lonchodomas jiantsaokouensis • Miraspis sp. • « Hispaniaspis sp. Remopleurides cf. pisiformis • Remopleurella burmeisteri • PALAEONTOLOGY, VOLUME 40 400 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Section through the upper part of the Pa Kae Formation at its type locality showing ranges of trilobites within the section. Base of the measured section is taken above massive crinoidal limestones forming the lower part of the formation. The upper fauna is identical in many respects to that of the Pagoda Limestone Formation of southern China (Lu 1975; Ji 1986; Sheng and Ji 1987; Zhou and Xiang 1993). A great majority of the whole list of species is in common, including a number of distinctive forms which are known nowhere else but in the Pagoda and Pa Kae formations. These latter include: Elongatanileus convexus , Remopleurella insculpta, Cyclopyge recurva. Parvigena plana , Hanjiangaspis fibrisulcata , Quyuania cf. ziguiensis , Panderia orbiculata , Hadromeros xiushanesis and Sphaerexochus fibri- sulcatus. Geographically more wide-ranging species, some of which have ranges extending into strata equivalent to the lower Ashgill, include Corrugatagnostus jiangshanensis, Nileus transversus, Remopleurella bunneisteri , Microparla speciosa, Telephina convexa , Paraphillipsinella globosa, P. nanjiangensis , Oedicybele sulcata , Ovalocephalus ovatus and Parisoceraurus rectangularis. Ji (1986, p. 8) noted that Ovalocephalus ovatus and both Paraphillipsinella species are present in all Pagoda Limestone Formation localities in Sichuan, Guizhou, Shaanxi, Hubei and Anhui provinces. There can be no doubt at all that the Pa Kae Formation and the Pagoda Limestone are correlative. At the southern edge of the northern China platform (Ordos Basin, Shaanxi Province) faunas comparable to those of the Pagoda Limestone Formation intergrade with graptolitic facies (Fu et al. 1993), and this indicates a correlation with the Climacograptus bicornis Biozone. Wang (in Chen et al. 1995, p. 64) cited evidence that the Pagoda Formation ‘ranges through the interval from the D. clingani Zone to the lowest part of the P. linearis Zone’, i.e. is Caradoc. However, Sheng and Ji (1987) concluded that the top of the Pagoda Formation may extend into the lower Ashgill. The Pagoda Limestone in China does not appear to have been formally divided biostrati- graphically, although, as described above, there is some evidence to suggest that two successive FORTEY: ORDOVICIAN TRILOBITES 401 table 1 . (Left) Comparison of species and genera in common with Chinese faunas (expressed as percentage of Pa Kae fauna), to show identity with the Pagoda Limestone Formation. For the purpose of this comparison, provisional determinations (cf., ?, etc) are regarded as conspecific. (Right) Dendrogram showing relationship of Pa Kae trilobitefauna to those from other areas as expressed in decreasing numbers of genera in common. Species (%) 15 Genera (%) 29 10 18 44 61 35 61 10 25 13 39 e,r ^ 20°), and any difference in the position of the eye between Cyamella stensioei and P. hujiabaensis is accounted for by the slightly longer palpebral lobe of the latter. The species from Thailand is effaced like Paracyamella but with the palpebral lobe like Cyamella stensioei. Hence the differences between the genera are both slight and intergradational, and I prefer to regard Cyamella as the senior synonym. Cyamella sp. 1 Plate 5, figures 6, 9 Material. Cranidium, It 25455. Stratigraphical range. Pa Kae Formation section, 144 m above base. Description. The well-preserved cranidium has a weakly tapering glabella with a truncate front, which shows only the faintest indications of lateral glabellar furrows of rorringtoniid form. The occipital furrow is better FORTEY: ORDOVICIAN TRILOBITES 425 defined; the occipital ring has a median tubercle. The preglabellar field slopes downwards to a rather ill- defined, almost concave, border. Anterior branches of the facial suture diverge at about 30° to the sagittal line as seen in dorsal view. Palpebral lobes are one-third length of glabella (sag.). Remarks. The poorly defined anterior cranidial border and the truncate glabella are the most obvious differences from C. stensioei. The closest species is C. hujiabaensis, from the Pagoda Limestone Formation (especially Zhou and Xiang 1993, pi. 4, fig. 1), which, however, has an evenly parabolic glabellar outline, and less divergent anterior branches of the facial sutures. The glabella outline resembles that of some Rorringtonia species, e.g. R. lenis Owens and Hammann, 1990. There is not enough material to name this species formally. Genus phaseolops Whittington, 1963 Type species. Phaseolops sepositus Whittington, Cow Head Group (middle Ordovician), western Newfoundland, by original designation. Phaseolops ? cf. conus Hu, 1971 Plate 5, figures 14-15 Material. Cranidium, It 25454. Stratigraphical range. Block from lower part of Pa Kae section, not precisely localized. Remarks. A single specimen is compared to a species described from silicified specimens by Hu (1971, p. 1 11) as Phaseolops conus from the Edinburg Formation of Virginia. I doubt whether Hu was correct in referring this species to Phaseolops. Whittington’s type species has a well-developed S2, as well as palpebral lobes placed close to the glabella in proetid fashion. P. conus , on the other hand, has a weak S2, and palpebral lobes with distinct rims, and well removed from glabella in rorringtoniid fashion. The Thai species is very like that from the Edinburg Formation, apart from having a narrower cranidial border. It even shows an unusual sculpture of scattered tubercles on the preglabellar field, which is also present on the American species. This is the one connection with faunas outside Gondwana, but a single specimen is not adequate for a reconsideration of P. conus. Here I place the species in Rorringtoniidae, and simply compare it with Hu’s species. Family telephinidae Marek, 1952 Genus telephina Marek, 1952 Type species. Teleplms fractus Barrande, 1852, Ashgill, Bohemia, by original designation. Telephina convexa Lit, 1975 Plate 5, figures 16-17; Text-figure 3 Synonymy. See Tripp et al. 1989, p. 44. Stratigraphical range. Upper part of the Pa Kae Formation section, 14-4-42 m above base. Material. Cranidia, It 25272, 25877; free cheeks. It 25269-70, 25429. Remarks. The description of the type material by Lu (1975) mentioned only cranidia, but Tripp et al. (1989) assigned a free cheek and pygidium. The well-preserved cranidium from the Pa Kae Formation illustrated here is exactly like Lu’s holotype, but shows the small, near vertical occipital 426 PALAEONTOLOGY, VOLUME 40 spine developed near the back end of the occipital ring which Tripp et al. (1989) recorded on material form the Tangtou Formation. Also typical of the species is very coarse tuberculation which does not extend on to the muscle insertion areas, in which the best Thai specimen matches the holotype and that illustrated by Lu and Zhou (1981, pi. 6, fig. 8); sculpture is weakly reflected on internal moulds. The glabellar muscle impressions are clearly shown; it should be noted that the bulk of the fixigenal field is another such area, so that the tuberculate sculpture is confined to its perimeter. A species described by Weir (1959) as Telephina cf. reedi from the Ashgill of Co. Clare, Ireland, has apparently similar sculpture; the rectangular shape of the glabella of the material illustrated is certainly the product of tectonic distortion. I have prepared the anterior cranidial border of the Thai cranidium, which is curved into an inverted TJ ’ (PI. 5, fig. 17). A species described as Telephina hangzhongensis from the Pagoda Limestone by Chen (in Li et al. 1975) shows a similar border and sculpture, and may be a synonym of T. convexa. The pygidium associated with this species (Chen in Li et al. 1975, pi. 19, fig. 2 a-b) is assuredly not that of a Telephina , being both too convex (tr.) and having too many pleural segments. The reconstruction of Telephina spinifera given in the Treatise (Whittington in Moore 1959, p. 298) is incorrect in portraying the border as a pair of anteriorly directed spines. The two large glabellar muscle impressions may indicate that the number of cephalic limbs was reduced in Telephina , possibly in connection with its pelagic mode of life. The square-lensed eyes (Text-fig. 3) are described above. Tripp et ah (1989) made no mention of anything unusual about the eye lenses. The immature cheek illustrated on their figure 9u seems to show polygonal lenses, and very few files dorsoventrally, and so it may be the case that in earlier instars the eye was of a more conventional holochroal type. At intermediate size the Thai material shows at least 25 lenses counting along an obliquely arranged row of ‘squares’ from bottom to top of the eye. Ji (1986) recorded T. convexa from the Pagoda Limestone, and, although the type material is from the slightly younger Linhsiang Formation (early Ashgill; see Chen et al. 1995), the similarity between type material and the new collections from Thailand indicates their conspecificity. Family phillipsinellidae Whittington, 1950 Remarks. Xia (1978, p. 176) erected a family Quyuaniaidae [sm] based on the monotypic genus Quyuania, which was erected in the same paper. I regard this genus as a phillipsinellid and, if this view is correct, a separate family is redundant. Genus paraphillipsinella Lu, in Lu and Chang, 1974 Type species. Paraphillipsinella globosa Lu, in Lu and Chang, 1974, Pagoda Formation (Caradoc), Sichuan, China, by original designation. EXPLANATION OF PLATE 6 Figs 1, 3-5, 9. Paraphillipsinella globosa Lu, 1974. 1, 3, 5, It 25879; cephalic shield in anterior, lateral and dorsal views; 18 m; x 15. 4, It 25880; cephalic shield with frontal lobe of glabella broken off, showing eyes; 18 m; x 13. 9, It 25881 ; incomplete cranidium; 0-6 m; x 12. Figs 2, 10. Paraphillipsinella nanjiangensis Lu, 1974. 2, It 25323; cast from mould of small cephalic shield; 41 m; x 12; 10, It 25353; cranidium; 39 m; x 18. Fig. 6. Quyuania cf. z iguiensis Xia, 1978; It 25453; cranidium; 14-4 m; x 9. Figs 7-8, 11-12, 15. Cekovia transversa (Ji, in Sheng and Ji, 1987). 7, It 25444; small cranidium; 18 m; x 10. 8, It 25288; larger pygidium with doublure; 18 m; x 5. 1 1, It 25326; pygidium; 18 m; x 10. 12, 15, It 25279; cranidium, dorsal and anterior views; 24 m; x 4. Figs 13-14, Cekovia striata Ji, 1986. 13, It 25382; larger pygidium, incomplete on right hand side; 37 m; x 8. 14, It 25882; transitory pygidium retaining one unreleased segment; 18 m; x 10. Specimen details as for Plate 1, figures 1, 3-12. PLATE 6 FORTEY, Ordovician trilobites 428 PALAEONTOLOGY, VOLUME 40 Paraphillipsinella nanjiangensis Lu, 1974 Plate 6, figures 2, 10 Synonymy. See Tripp el al. 1989, p. 43. Material. Cephalic shields and cranidia. It 25246-25249, 25323, 25353, 25356 a-b. Horizon. Upper part of Pa Kae Formation section, 37-42 m above base. Remarks. There is uncertainty about how many species of Paraphillipsinella may be recognized. Rather a large number has now been described from the Yangtze region of southern China, but their range of variation is unclear. A complete exoskeleton was figured by Qiu et al. (1983, pi. 76, fig. 9), but there is only cephalic material from Thailand. Ji (1982) reviewed the genus and gave some measurable parameters with which to distinguish up to seven species. Lu and Zhou (1981) recognized only two species: P. globosa Lu, in Lu and Chang, 1974, with a sub-circular frontal glabellar lobe, and P. nanjiangensis Lu, in Lu and Chang, 1974, with a more transversely oval frontal lobe. The Thai species illustrated in Plate 6, figure 2 is clearly different from P. globosa (Lu in Lu and Chang 1974, pi. 53, figs 8-9; Zhou and Dean 1986, pi. 62, figs 13-16), and more like P. nanjiangensis with regard to the frontal glabellar lobe. Ji (1982, fig. 2) recognized P. nanjiangensis as the stratigraphically earlier part of a lineage ranging through the Pagoda Limestone and comprising this species and P. funga Ji, 1982. However, because Tripp et al. (1989) recognized P. nanjiangensis from the formation above the Pagoda Limestone the stratigraphical difference is unlikely to apply. The choice of assignment of the Thai species is between these two species, if indeed they are distinct. The ratio of maximum to minimum glabella width given in Ji’s (1982, p. 60) chart appeared to provide a distinction between nanjiangensis and funga , the latter being relatively broader. However, the ratio in well-preserved Thai material spans the range (2-9 to 3-5) for the two species. I think it likely that these two taxa are just part of a single variable species. Tripp et al. (1989) thought that funga might have a less inflated glabella, a difference I find hard to apply given the range in preservation styles of the material illustrated in previous works. Accordingly, I use P. nanjiangensis here. The cephalic material from Thailand is beautifully preserved; one specimen (PI. 6, fig. 2) retains the free cheek showing the indistinct eye lobe and short genal spine. Paraphillipsinella globosa Lu, in Lu and Chang, 1974 Plate 6, figures 1, 3-5, 9 Synonymy. See Zhou and Dean (1986, p. 767). Material. Cephalic shields. It 25879-25880; cranidia. It 25451-25452, 25496-25497, 25881, 25903-25906; pygidia. It 25498, 25908-25909. Stratigraphical range. Lower to middle part of Pa Kae Formation, 0-6-18 m above base. Remarks. This species has been described several times, most recently in English by Zhou and Dean (1986), who itemized several synonyms in the Chinese literature. The almost perfectly spherical frontal glabellar lobe immediately differentiates this species from P. nanjiangensis. The small, nearly entire cephalic shield illustrated here is probably the best preserved specimen known. It shows the eye surface very well, but even using high magnification I have been unable to see any lenses (which are commonly seen on other Thai species) and I presume they were minute. The venter bulges downwards (PI. 6, fig. 3) so that the frontal glabellar lobe is essentially a sphere; raised lines on the genal border pass on to the frontal lobe where they split into anastomosing lines, and ventrally there are strong cuesta-like terrace ridges of which the scarp slopes face anteriorly. FORTEY: ORDOVICIAN TRILOBITES 429 Genus quyuania Xia, 1978 Type species. Quyuania ziguiensis Xia, 1978, Pagoda Limestone, by original designation. Remarks. Although accommodated in a separate family by Xia (1978), the type species of Quyuania appears similar to some species of the matutina species group which have been assigned to Phillipsinella (Bruton 1976, p. 707), for example P. fornebuensis Bruton, 1976, from the Caradoc of Norway. In comparison with Phillipsinella parabola , these species have a relatively unexpanded (tr.) frontal glabellar lobe. Quyuania may be retained for those species in which the maximum transverse glabellar width is less than twice the minimum width of the glabella behind, and which have an occipital ring which is not unusually long (sag.). Tripp (1962) established a genus, Kirkdomina, on the basis of small specimens of the type species, K. williamsi, from the Confinis Flags (Llanvirn) of the Girvan District, Scotland. It is similar to Quyuania apart from in its narrow (tr.) anterior cranidial border, but I am uncertain whether the similarity is significant in view of the immaturity of Tripp’s specimens. Quyuania cf. ziguiensis Xia, 1978 Plate 6, figure 6 Material. Cranidium, It 25453. Occurrence. Pa Kae Formation section, 14-4 m above base. Remarks. Q. ziguiensis from the Pagoda Formation has been figured by Xia (1978), Ji (1986) and Sheng and Ji (1987), and has a distinctive sculpture, comprising bowed, elevated lines on the glabellar frontal lobe, which can be matched on the Thai cranidium. However, the frontal glabellar lobe is even less inflated on the latter, and the cranidial border may also be wider. If it is a different species, the specimen is not an adequate basis for naming it formally. Family illaenidae Hawle and Corda, 1847 Genus cekovia Snajdr, 1956 Type species. Illaenus transfuga Barrande, 1852, Caradoc, Bohemia, by original designation. Remarks. The differences between certain species of Parillaenus and those of Cekovia are not great. For example, Parillaenus dalecarlicus (Warburg, 1925) was redescribed by Bruton and Owen (1988), and has a comparatively well-defined, waisted glabella, unlike the type species, P.fallax Holm, but like some Cekovia species. The species from Thailand has a similar glabellar shape. Since this is likely to be a plesiomorphic (styginid-form) character, its significance in generic diagnosis is unclear. However, a well-defined pygidial axis seems to be more characteristic of Cekovia than of typical illaenids. Hammann (1992, p. 62) cited the Pagoda species C. striata Ji, 1986, as a typical Cekovia and the species described below is referred to the same genus. *Cekovia transversa (Ji, in Sheng and Ji, 1987) Plate 6, figures 7-8, 11-12, 15 1987 Illaenus transversus Ji, in Sheng and Ji, p. 30, pi. 1, figs 1-2. Material. Cranidia, It 25279, 25440, 25444; pygidia. It 25288, 25326, 25487. Stratigraphical range. Pa Kae Formation section, 0-6-24 m above base. See note on p. 449. 430 PALAEONTOLOGY, VOLUME 40 Description. The material from Thailand is rather better preserved than the specimens from the Pagoda Limestone used by Ji to found Illaenus transversus, and such differences as there are may be no more than preservational. On Thai material, the cranidium is moderately convex, and the glabella is distinctly waisted at one-quarter to one-third cranidial length in dorsal view. At its narrowest, the glabella is half the cranidial length. The smaller cranidium shows a well-marked occipital tubercle. Palpebral lobes are large for the genus, approximately one-third cranidial length, but I cannot be certain of this feature on Ji’s material. Surface sculpture of terrace ridges on the anterior part of the cranidium where the axial furrows are obsolete. Distinctive pygidium with length/width ratio ranges from 04 to 06, and the axis occupies one-quarter of the transverse width, and about one-third the length. The axis is wider than long with an acutely triangular outline. Two narrow and obscurely defined axial rings are visible on smaller specimens, but not on larger ones. There is a faint median postaxial ridge. There is a pair of outwardly diverging furrows running across the pleural fields from the mid length of axial furrows; these extend as far as the paradoublural line, and may represent its adaxial continuation. The doublure on the largest pygidium proved difficult to prepare, but certainly widens mesially (PI. 6, fig. 8). Remarks. Despite its distinctive pygidium, this is a difficult species to name with certainty. Ji (in Sheng and Ji 1987) illustrated a cranidium and pygidium, rather indistinctly. However, the pygidium shows the outward curving furrows crossing the adaxial parts of the pleural fields which are a distinctive feature of the species. Ji’s specimens are larger than the Pa Kae specimens, which may account for their wider cephalic and pygidial axes. However, it would be difficult to apply a different name in view of the similarity of the pygidia, given the identity of other Thai species to those from the Pagoda Formation, ft is unclear whether Ji’s designation is valid, because there is no formal description, apart from the legend to the figure. The Chinese cranidium illustrated is very much larger than any discovered from the Pa Kae Formation, which may account for its greater effacement and smaller eyes. Other details are uncertain because of the poor illustration. The Thai cranidium is very like that attributed by Ji (1986) to Cekovia striata from the Pagoda Limestone Formation; however, the pygidium assigned to that species lacks the characteristic dorsal furrows, and is longer (sag.). I have assigned the cranidia figured here to transversa because they are associated in one bed in similar abundance. Cekovia striata Ji, 1986 Plate 6, figures 13-14 Material. Pygidia, It 25382, 25882-25883; ?cranidium. It 25383. Stratigraphical range. Upper part of Pa Kae Formation section, 18-42 m above base. Remarks. Pygidia differ from those attributed to C. transversa in lacking the distinctive dorsal furrows of that species, and in having a more concave-sided pygidial axis. Similar differences apply to C. striata , to which the Thai species is assigned. An immature pygidium figured herein is similar to that assigned by Ji (1986, pi. 3, fig. 12) to Cekovia striata in showing a post-axial ridge. Family panderiidae Bruton, 1968 Genus panderia Volborth, 1863 Type species. Panderia triquetra Volborth, 1863, lower middle Ordovician of St Petersburg, by monotypy. Panderia orbiculata Ji, 1986 Plate 7, figures 1-6 1986 Panderia orbiculata Ji, p. 17, pi. 4, figs 3-6. 1987 Panderia orbiculata Ji; Ji, in Sheng and Ji, pi. 1, fig. 18. FORTEY: ORDOVICIAN TRILOBITES 431 Material. Cephala, It 25307-25308; cranidia. It 25309, 25235-25236; pygidia. It 25233, 25411, 25428; cheeks and genal doublure. It 25237, 25306. Stratigraphical range. Upper part of Pa Kae Formation section, 18-42 m above base. Description. Ji’s short description of P. orbiculata from the Pagoda Limestone is supplemented here. P. orbiculata belongs within a section of Panderia with comparatively well-defined axial furrows; Ji’s specimens are exfoliated, but furrows are also distinct on the material from Thailand, which retains its cuticle. Bruton (1968, p. 2) has pointed out that illustrations of Panderia cranidia vary greatly according to how the specimens are oriented. Dorsal and palpebral views are used here according to Bruton’s definition. In palpebral view, the cephalon is two-thirds as long as wide. In dorsal and palpebral views the axial furrows diverge outwards- forwards, converging anteriorly. Glabella distinctly convex ( tr. ) ; in the best-preserved specimen the anterior part is steeply downsloping and the posterior is close to horizontal, and there is a rather sharp break in slope between these two parts, producing a blunt point on the profile (PI. 7, fig. 3). The same specimen shows a pair of prominent muscle insertion areas opposite the posterior part of the eye, and two small and faint pairs behind. Eyes approach half cephalic length in palpebral view; in lateral view the eye is about three times as long as high. Eye lenses are exceedingly small and numerous. Free cheeks are very narrow (tr.), without border. The distance from the front of the eye to the anterior cephalic margin is short, and the anterior section of the facial suture reflects this. A pygidium (PI. 7, fig. 5) is typical of the genus and is probably correctly associated. The axis is well defined for the genus, and the doublure outline is parallel to the posterior pygidial margin. Remarks. This species is very like an Ashgill species from Norway, P. insulana Bruton, 1968, which also has the eye extending far forwards. It differs from this, and other Panderia species, in the low downward curvature of the posterior part of the glabella, such that in dorsal view the cranidium is relatively long (sag.). The eye is proportionately deeper in P. orbiculata ; on P. insulana the visual surface is at least four times longer than deep. Although the Thai material is better preserved than that illustrated by Ji ( 1986, pi. 4, fig. 6), the cranidium of the specimen illustrated on Plate 7, figure 6 is identical, and it seems very likely that it is the same species. The poorly illustrated Panderia sp. of Apollonov (1974) from the upper Ordovician of Kazakhstan may prove to be the same species. Panderia migratoria Bruton, 1968 Plate 7, figures 7-14 Material Cephala, It 25854-25855, 25857; cranidia. It 25180, 25563-25566, 26202; pygidia. It 25345, 25856, 25866; free cheek. It 25562. Stratigraphical range. Lower part of Pa Kae Formation section, 0-6-39 m above base. Remarks. Bruton ( 1968) gave a full description of this species from Caradoc occurrences in Norway and Sweden. It is an extremely convex form, and thus easily distinguished from P. orbiculata. I can find no important differences between the Thai and Scandinavian material. This is a highly effaced member of the genus, and it is possible that the Thai specimens are more so than the type series since the axial furrows are faint, even at the posterior end of the glabella, but this would scarcely be a reliable specific distinction. Of the species described by Bruton (1968) only P. edita is as convex, but this is a species having a long anterior branch of the facial suture. Family staurocephalidae Prantl and Pribyl, 1948 Genus oedicybele Whittington, 1938 Type species. O. kingi Whittington, 1938, Ashgill, North Wales, by original designation. Remarks. The type material of the type species of Oedicybele Whittington, 1938 is not well preserved but Kielan (1957) figured better material from Poland, which she attributed to O. kingi. 432 PALAEONTOLOGY, VOLUME 40 This species has minute, anteriorly positioned eyes, and may be regarded as an example of an atheloptic (Fortey and Owens 1987) trilobite in which the eyes are reduced in function. Oedicybele kildarensis Temple, 1965 described from the Ashgill of Ireland (Temple 1965; Dean 1971), from good material, has somewhat larger, but still small, eyes. The type species of Dindymenella Lu (in Lu et al. 1976), D. sulcata , from the upper Ordovician of Yunnan, is blind, and is known from poorly preserved type material. In other features, however, it is extremely like Oedicybele. If I am correct in identifying the material from Thailand with the species sulcata , this affords a good basis for comparison with the better preserved material of Oedicybele. Glabellar structure is identical, not only in the presence of basal bacculae-like glabellar lobes, but also in the form of the lateral glabellar furrows, especially a shallow S4 running parallel to the axial furrow. Other details, such as the convexity of the cheeks, and even the surface sculpture, are also closely comparable. Note that the glabellar structure of the Thai species is also quite different from that of Dindymene and allied genera, with which Dindymenella might otherwise by compared on account of its blindness. The development of eyes in atheloptic trilobites is variable, and no particular taxonomic significance can be attached to the presence of relict eyes as compared with their complete absence. Thus, for example, within the dalmanitoidean genus Ormathops there are species with small anterior eyes and others with no eyes at all (Fortey and Owens 1987). Hence the grounds for erecting Dindymenella seem insufficient, and it is here considered a subjective synonym of Oedicybele. Oedicybele sulcata (Lu, in Lu et al., 1976) Plate 10, figures 7, 10-11 Material. Cephalic shield. It 25887. Stratigraphical range. Uppermost part of the Pa Kae Formation section, 42 m above base. Description. Cephalon twice as wide as long in dorsal view, convex forwards. Glabella also protrudes forwards, and expands greatly over its anterior two-thirds, such that the width at the occipital ring is half that of the frontal lobe. Basal bacculae-like glabellar lobes are conspicuous, as they are in O. kingi (Kielan, 1957, pi. 6, fig. 3) although not shown in Kielan’s reconstruction of this species. S1-S3 are short, deep, more or less transverse pit-like furrows, and S3 is continued very faintly towards the mid-line of the glabella. S4 is shallower, running close and parallel to axial furrow. Axial furrows deep, but narrow. Convex fixed cheeks show no sign of eye or facial suture. Borders widen towards genal angle where there is a short genal spine directed somewhat outwards. Narrow posterior border furrow curves round into lateral border. Dorsal cuticular surface with scattered, large tubercles, noticeable especially on glabellar frontal lobe. Genal prosopon finely reticulate. Remarks. This species differs from O. kildarensis Temple in having genal spines and a finer scale reticulate sculpture on the fixed cheeks. It resembles O. kingi Whittington in the same features; however, in the specimen of O. kingi figured by Kielan (1957, pi. 5) the sculpture extends on to the glabella also. O. kildarensis has the largest eyes. On O. kingi they are much reduced, and on EXPLANATION OF PLATE 7 Figs 1-6. Panderia orbiculata Ji, 1986. I, 3, It 25307; cephalon, in dorsal and lateral views; 37 m; x 10. 2, 6, It 25236; cranidium, in lateral and palpebral views; 42 m; x 9. 4, It 25308; small cephalon; 37 m; x 10. 5, It 25428; pygidium; 42 m; x 7. Figs 7-14. Panderia migratoria Bruton, 1968. 7, 10-11, It 25854; cephalon, in palpebral, lateral, and dorsal views; x9. 8, 12, 14, It 25855; smaller cephalon in palpebral, dorsal and lateral views; x9. 13, It 25856; pygidium; x 13. 9, It 25857; cephalon, anterior view; x 9. All 18 m. Specimen details as for Plate 1, figures 1, 3-12. PLATE 7 FORTEY, Pcmderia 434 PALAEONTOLOGY, VOLUME 40 O. sulcata they are lost altogether. Lu’s original specimens are more complete but less well preserved than the Thai material, and one cannot compare sculptural details, but Lu states that the species is without eyes, and hence the closest comparison of this cephalon is with O. sulcata rather than O. kingi. A cranidial fragment from the Pagoda Limestone Formation figured by Ji (1986, pi. 6, fig. 10) as Atractopygel sp. appears to be similar. Oedicybele sp. nov. A Plate 10, figures 8-9 Material. Cranidium, It 25477; another unnumbered cranidial fragment. Stratigraphical range. Lower part of Pa Kae Formation section, 0-6-18 m above base. Description. The single well-preserved cranidium is not an adequate basis on which to name what is likely to be a new species. In dorsal view the tumid glabella expands evenly forwards such that its transverse width across the frontal lobe is twice that immediately in front of the occipital ring. Glabellar furrows deep, but short. SI transverse, showing a deep exterior part but also continues very faintly across mid-part of glabella. S2 pit-like, directed slightly forwards and inwards. S3 present only on the flanks of the glabella, a shallow pit. The frontal glabellar lobe carries large and scattered low, round tubercles. Posterior parts of fixed cheeks are strongly convex upwards. Posterior border narrow close to glabella and widening rather abruptly near genal angle. I have not succeeded in preparing the palpebral lobe although the course of the facial suture clearly shows that there were free cheeks present of about the same size as those in O. kildarensis (Dean 1971, pi. 17, fig. 6). Remarks. The species from Thailand displays the same rounded and scattered glabellar tubercles as O. kildarensis Temple, 1965, from the Chair of Kildare Limestone (Ashgill), Ireland, which was well described by Dean (1971). However, O. kildarensis is consistently different in having the glabellar furrows strongly incised across the median lobe, such that the frontal glabellar lobe is isolated from the posterior part of the glabella. In both O. kildarensis and O. kingi (see Kielan 1957) the axial furrows diverge forwards more strongly around the frontal lobe than they do in the Thai form, so that the outline of the furrows is curved in these species. The glabella presumably represents the plesiomorphic condition in which the furrows are more Atractopyge- like. Family hammatocnemidae Kielan, 1960 Genus ovalocephalus Koroleva, 1959a Type species. Ovalocephalus kelleri Koroleva, 1959a, Caradoc of Kazakhstan, by original designation. Remarks. Zhou and Dean (1986, p. 776) noted that Ovalocephalus was likely to be the senior synonym of Hammatocnemis Kielan, 1960. Through the kindness of Dr Koroleva I have been EXPLANATION OF PLATE 8 Figs 1-10. Ovalocephalus plewesae sp. nov. 1, 5-6, holotype. It 25518; cranidium, in dorsal, lateral and anterior views; x 10. 2, It 25522; incomplete cranidium, showing shallow S2; x 10. 3-4, It 25521; hypostome in ventral, x 12, and lateral, x 8, views, the latter showing wings. 8, It 25520; small cranidium; x 15. 7, 9, It 25524; pygidium, dorsal and lateral views; x 16. 10, It 25523; plan view of small, incomplete free cheek; x 10. All 18 m. Fig. 11. Josephulus gracilis Warburg, 1925; cast of holotype, RM D-188, Swedish Museum of Natural History; cranidium; Boda Limestone (Ashgill), Sweden; x 5. Fig. 12. Parisoceraurus rectangularis Zhou, 1977; It 25430; cranidium; 1 4-4 m; x 7. Specimen details of figs 1-10, 12 as for Plate 1, figures 1, 3-12. PLATE 8 FORTEY, Ordovician trilobites 436 PALAEONTOLOGY, VOLUME 40 supplied with a photograph of the holotype of the type species, which is illustrated as Plate 9, figure 7 herein. These confirm the synonymy; the anterior glabellar furrows are more effaced on the type species of Ovalocephalus than on the type species of Hammatocnemis , H. tetrcisalcatus Kielan, 1960, but this is a variable feature in the group and not of generic significance. Another genus which may belong in Hammatocnemidae is Josephulus Warburg, 1925; the type species, J. gracilis, is illustrated here (PI. 8, fig. 11) from its cranidium, all that is described of it. It shares the distinctive glabellar shape of Ovalocephalus tetrasulcatus. It differs from all the species attributed to Ovalocephalus in having a narrow (sag.) occipital ring, long genal spines and a clearly distinct lateral border on the cranidium. Ovalocephalus plexvesae sp. nov. Plate 8, figures 1-10 1988 Ovalocephalus kel/eri Koroleva; Dean and Zhou, p. 776, pi. 64, figs 13-14 [sub Hammatocnemis kelleri [sic]]. Derivation of name. For Caryl Plewes, who helped the author with preparation of material. Holotype. Cranidium, It 25518. Paratvpes. Cranidia, It 25520, 25522, 25476; hypostome. It 25521; free cheek, It 25523; pygidium. It 25524. Stratigraphical range. Lower part of the Pa Kae Formation section, 0-6—1 8-0 m above base. Diagnosis. Ovalocephalus having glabella produced into an anterior, spine-like ‘nose’; S3 and S4 glabellar furrows effaced; surface sculpture lacking; genal spine present. Description. The cranidium of this species lacks surface sculpture. All other Ovalocephalus species are granulose. Hence it is likely that the pygidium, free cheek and hypostome, are correctly assigned, because they are from the same horizon as the type cranidium, and also lack surface sculpture. Mature cranidium longer than wide. Much of this length is accounted for by the extension of the glabella into a long ‘nose’ anteriorly. On the small cranidium this extension is less pronounced, but a spine-like protuberance is clear. The glabella is less inflated than it is in other species of Ovalocephalus. Glabella tapers forwards to a point behind S2, then expands forwards to a maximum width which is about twice minimum width. Axial furrows are narrow. Glabellar furrows are also narrow, but distinct. Only two pairs are incised; SI slopes inwards and backwards to approach the occipital furrow closely; S2 is shorter and is directed slightly anteriorly. Hence, the two furrows enclose an acute angle. There is a comparatively weak furrow running across the glabella connecting the inner ends of SI. A very faint third pair of glabellar furrows, isolated within the glabella, is shown by the larger specimen on Plate 8, figure 2. The occipital ring is very wide (sag.) and widest medially. The occipital furrow is narrow, and of equal depth along its length. The fixed cheeks are gently convex, more or less continuing the downward slope of the anterolateral part of the glabella. The posterior fixigenal border has about half the width (exsag.) of the occipital ring. Short, stout, pointed genal spine present. The free cheek (eye not preserved) has a smooth genal field, but a highly convex border carrying a few raised lines. A hypostome (PI. 8, figs 3-4) is similar to those (but perhaps better preserved) which have been associated with other species of Ovalocephalus (Lu and Zhou 1979, pi. 3, fig. 6; Dean and Zhou 1988, p. 57). Middle body gently convex with a pair of slightly depressed maculae weakly indicated at its mid-length, well isolated from the marginal furrows. Borders narrow and convex, widening gently backwards, and distinctly defined by narrow border furrows. Posterior margin carries a pair of posterolateral spines, and a median one is indicated. The posterior thoracic segment is attached to the pygidium, showing gently convex axis narrower (tr.) than pleurae. Pleura divided into adaxial horizontal part, and lateral downsloping part with bluntly spinose termination. The adaxial part carries narrow (exsag.) posterior articulating ridge, which engages with boss on front of pygidium. Pygidium more than twice as wide as long, broadly arched upward about mid-line. Axis tapers posteriorly, but its posterior end is not defined. The four axial rings are progressively shorter backwards (sag.) and the axial rings defining them also curve more markedly forwards medially. There are three pleural ribs: the first is completely defined, reaching the margin; the furrow defining the second does not curve backwards at its outer FORTEY: ORDOVICIAN TRILOBITES 437 end, where it fades out before reaching the margin, the furrow defining the third is extremely short, present adaxially only. Remarks. The type species, O. kelleri Koroleva, 1959a (see also Koroleva 19596), has a less pointed frontal glabellar lobe than O. plewesae, and also a granulose surface sculpture, and its basal glabellar lobes are sub-circular and inflated. Chinese species attributed to this genus were reviewed by Lu and Zhou (1979), and none of these has the glabellar ‘nose’ of the new species, the S3 and S4 furrows are developed, and, where the cuticle is preserved, a granulose sculpture is general. The same distinctions apply to O. hexianensis (Q. Z. Zhang in Qiu et al. 1983) and O. tetrasulcatus (Kielan, 1960). Where genal spines are present at all on these species they are very short and stubby. The closest comparisons with O. plewesae are with species figured by Zhou and Dean (1986) from the Chedao Formation, Gansu. Their species, H. obsoletus , is likewise smooth, and the anterior glabellar furrows are similarly obsolete. However, its glabellar frontal lobe is rounded, and a genal spine is lacking from much wider (tr.) fixed cheeks, and it seems very unlikely that such consistent and prominent differences could be intraspecific. The pygidium attributed to O. obsoletus is much like that of O. plewesae. Zhou and Dean also recognized what they identified as O. kelleri Koroleva. The specimen on their plate 64, figures 13-14 is very like O. plewesae , and unlike O. kelleri both with regard to the outline of the glabellar frontal lobe and in having genal spines; the basal glabellar lobes are a little more inflated than in the Thai material and the furrow joining their inner ends is stronger, but the similarities are sufficient to indicate that they should be regarded as conspecific. The Gansu species is Caradoc in age. Ovalocephalus ova t us (Sheng, 1964) Plate 9, figures 1-6, 8, 12 1964 Hammatocnemis tetrasulcatus var. ovatus Sheng, p. 560, pi. 2, fig. 2 a-c. 1975 Hammatocnemis ovatus Sheng; Lu, p. 441 (231), pi. 45, figs 1-3. 1975 Hammatocnemis pagoda Chen (in Li et al.), pi. 21, figs 7-8. 1977 Hammatocnemis ovatus Sheng; Wang and Jin (eds), p. 252, pi. 76, fig. 6. 1978 Hammatocnemis ovatus Sheng; Xia, p. 181, pi. 36, figs 6-7. 1986 Hammatocnemis ovatus Sheng; Xiang and Ji, p. 60, pi. 2, figs 13-14. 1986 Hammatocnemis ovatus Sheng; Ji, pi. 6, fig. 16. Material. Cephalon, It 25290; cranidia, It 25178, 25223-25224, 25320-25321, 25398-25399, 25401, 25403, 25432-25434, 25474, 25526-25527, 25536-25544, 25519, 25891, 25893-25897; pygidia. It 25222, 25338, 25545, 25898-25899, 26203-26205; free cheeks. It 25221, 25291, 25380, 25405, 25546, 25548; hypostomes. It 25225, 25341, 25549, 26206; thoracic segments. It 25525, 25547, 25550. Stratigrapliical range. Throughout measured section of Pa Kae Formation. Description. Thai material of this species is particularly well preserved, possibly the best for any species of the genus, and a description is therefore worthwhile. A description of the type material from the upper part of the Pagoda Formation was given by Sheng (1964), but his material was not well preserved. A fuller description by Lu (1975) cites the ovate anterior part of the glabella with convex-outward axial furrows as distinguishing O. ovatus from other species of the genus, features clearly seen on the Thai material. Lu’s specimens are at least partly internal moulds, whereas the Thai material retains the exoskeleton. The glabellar furrows thus appear a little shorter and narrower. Four pairs are preserved, of which the first, transglabellar furrow is by far the deepest and the three anterior pairs are more or less evenly spaced along the axial furrows and very short. The anterior pair is shortest and is almost obsolete on some specimens. The deep axial furrows enclose an angle of 60-70°; they diverge in front of the circular basal glabellar lobes. The frontal lobe of the glabella considerably overhangs the very narrow anterior border. In dorsal view, the length of the glabella in front of SI is equal to the maximum glabellar width in front of the palpebral lobes. The fragility of the limestone does not allow preparation of the ventral surface of the cephalon. The posterior border widens into a genal spine remnant. This is of interest because it shows that the suture can be considered homologically proparian (see 438 PALAEONTOLOGY, VOLUME 40 also Hammann 1992, p. 106), even though the posterior branch of the facial suture cuts the posterior margin of the cephalon immediately outside the genal spine (PI. 9, fig. 5). The genus was described as opisthoparian by Lu and Zhou (1979, p. 430) and Han (1980). As described above, the genal spine of O. plewesae is longer, and is also present on the cranidium. Lu and Zhou took the form of the facial suture as the defining character, because it reaches the posterior margin as in other opisthoparian trilobites. The species they described with cheeks in place lack noticeable genal spines. However, the recognition of the genal spines shows that in this unusual trilobite the posterior branch of the facial suture has shifted backwards. The alternative would be that the spines on Ovalocephalus were not homologous with ‘normal’ genal spines. While this is a possibility, it sits ill with the idea that Ovalocephalus is related to the proparian Cheiruridae, which Lu and Zhou (1979) argued from other grounds. The condition of the suture might be described as pseudo-opisthoparian. The eye is convex, deep, and placed rather far forwards; in the cephalic orientation with the occipital ring horizontal it is inclined forwards. The eye lenses appear to be normal holochroal. The eye is elevated on a convex eye socle, which carries a few larger tubercles. The lateral border is steeply sloping and bevelled. Surface sculpture is generally finely granulose other than on borders, but with scattered larger tubercles on frontal lobe, on genae, and on the eye socle. The lateral border carries a few raised lines. Four conjoined thoracic segments illustrate the high thoracic convexity and the cheirurid-like articulation and pleural spines. The pygidium is variable, up to three times as wide as long; terminal part of gently tapering axis effaced, anterior two rings clearly defined, a third may be well-developed, and a fourth faint on poorly preserved material. Similarly, the anterior three pairs of pleural spines are well differentiated but the third is often only indicated proximally. Remarks. This species closely resembles the type species, O. kelleri , from the Caradoc of Kazakhstan (Koroleva 1959a, 19596; Apollonov 1974, pi. 13, fig. 9; PI. 9, fig. 7 herein). There are two points of distinction: the glabellar furrows anterior to the second pair are effaced in O. kelleri , and the divergence of the axial furrows in front of SI is even lower; they enclose an angle of about 40°. This results in a longer frontal glabellar lobe, the length of the glabella in front of SI exceeding its maximum transverse width. Hammann (1992, pi. 22, fig. 12) figured a cranidium from the Cystoid Limestone of Spain very like that of O. ovatus under the name O. cf. tetrasulcatus. O. tetrasulcatus (Kielan) (see for example, Kielan 1960, pi. 25, fig. 3; pi. 26, figs 2-\\ pi. 27, figs 6-8; Apollonov 1974, pi. 13, figs 1-8; Lu and Zhou 1979, pi. 2, figs 10-11; Dean and Zhou 1988, pi. 59 figs 10, 12-16), the type species of Hammatocnemis , is also very similar, and is distinguished primarily by having angulate anterolateral corners of the glabella, which gives the anterior part of the glabella a sub-pentagonal outline in many specimens. I am uncertain if the apparently better definition of the glabellar furrows of tetrasulcatus might be attributable to differing preservation states. The better preserved of Kielan’s specimens (1960, pi. 26, fig. 4) show a wider divergence of the axial furrows, up to 90°, but this is not the case with specimens such as that illustrated by Lu and Zhou (1979, pi. 2, fig. 10). There is apparently no genal spine remnant on tetrasulcatus. Also generally similar to O. ovatus is O. decorosus Lu (in Lu and Chang, 1974) (see Lu and Zhou 1979; Tripp et al. 1989), a species having a less forwardly bulbous glabella, and apparently only a third pair of well-defined pygidial pleural ribs. Some pygidia attributed to tetrasulcatus by Apollonov (1974, pi. 14, figs 1-6) also show clearly defined third ribs. EXPLANATION OF PLATE 9 Figs 1-6, 8, 12. Ovalocephalus ovatus (Sheng. 1964). 1-2, 12, It 25474; well-preserved cranidium in dorsal, lateral and anterior views; 10-6 m; x 6. 3, It 25891 ; cranidium; 18 m; x 10. 4-5, It 25290; small cephalon, anterior and lateral views, the latter showing genal spine relative to facial suture; 18 m; x8. 6, It 25338; pygidium; 39 m; x 5. 8, It 25291 ; free cheek, lateral view showing sculpture and eye; 18 m; x 12. Fig. 7. Ovalocephalus kelleri Koroleva, 1959; holotype, Almaty Museum, Geological Institute, type collection; cranidium; Caradoc, Kazakhstan; x 3. Photograph kindly supplied by M. N. Koroleva. Figs 9-11. Hadromeros xiushanensis (Sheng, 1964). 9, It 25858; incomplete cranidium; 18 m; x 5. 10-11, It 25328; pygidium, dorsal and anterior views, showing angle of elevation of spines; 39 m; x 10. Except for figure 7, details as for Plate 1, figures 1, 3-12. PLATE 9 FORTEY, Ovalocephalus , Hadromeros 440 PALAEONTOLOGY, VOLUME 40 Family cheiruridae Hawle and Corda, 1847 Subfamily cheirurinae Hawle and Corda, 1847 Genus hadromeros Lane, 1971 Type species. Cheirwus keislevensis Reed, 1896, Ashgill, Cumbria, England, by original designation. Remarks. The phylogenetics of the cheirurids are not well understood. Lane (1971, text-fig. 10) showed an evolutionary scheme which derived Hadromeros from Xylabiorr, the differences between these two genera as listed by Lane (1971, p. 24) are mostly plesiomorphic characters of the latter (parallel-sided glabella with relatively low convexity) which do not provide a satisfactory diagnosis. Late Ordovician cheirurines are often referred to Hadromeros , to which one species from Thailand is cautiously assigned. However, there is another genus, Parisoceraurus Zhou (in Wang and Jin, 1977) (see also Tripp et al. 1989), which is stated to differ from Hadromeros in having a cranidial border (like Xylabion), eyes somewhat farther forward, and exceptionally long ‘great spines’ on the pygidium (like Ceraurus). There is a narrow cranidial border on the species described below, and Zhou (in Wang and Jin 1977) described two species of Parisoceraurus which are not different from Hadromeros xiushanensis in this feature, and Zhou evidently wished to assign this species also to his new genus. Clearly, the whole group needs critical revision, but I cannot attempt this on the basis of the small amount of material available from Thailand. Hadromeros xiushanensis (Sheng, 1964) Plate 9, figures 9-1 1 1964 Eccoptochile xiushanensis Sheng, p. 548, pi. 3, fig. 1 a-c. 1975 Paraceraurus sinicus Lu, p. 424, pi. 42, figs 9-10. 1975 Paraceraurus longisu/catus Lu, p. 425, pi. 42, fig. 11. 1986 Hadromeros xiushanensis (Sheng); Ji, p. 22, pi. 6, figs 8-9 (Inon figs 6-7). Materials. Cranidium, It 25858; pygidium. It 25328. Stratigraphical range. Upper part of Pa Kae Formation section, 18-39 m above base. Remarks. Ji (1986) synonymized several species described from the Pagoda limestone in H. xiushanensis. An incomplete cranidium from Thailand shows straight, long, and very deep glabellar furrows considered typical of the species. I have also prepared a pygidium with elongate anterior spines and reduced median ones which P. D. Lane informs me (pers. comm. 1995) is typical of EXPLANATION OF PLATE 10 Fig. 1. Miraspis sp.; It 25888; cranidium; 18 m; x 10. Figs 2-3, 5. Sphaerexochus fibrisulcatus Lu, 1975. 2-3, It 25494; cranidium, dorsal and anterior views; x 15. 5, It 25495; damaged cranidium; x 12. Both 42 m. Figs 4. 6. Hispaniaspisl sp. indet. 4, It 25886; incomplete free cheek, lateral view; x 20. 6, It 25892; incomplete cranidium; x 20. Both 18 m. Figs 8-9. Oedicybele. sp. nov. A; It 25477; cranidium, dorsal and anterior views; 0 6 m; x 8. Figs 7, 10-11. Oedicybele sulcata (Lu, in Lu et al., 1976); It 25887; cranidium, lateral, dorsal and anterior views: 42 m; x 12. Fig. 12. Lichasl sp.; It 25443; hypostome incomplete on right side; 14-4 m; x 6. Specimen details as for Plate 1, figures I, 3-12. PLATE 10 ' FORTEY, Ordovician trilohites 442 PALAEONTOLOGY, VOLUME 40 Hadromeros , and it seems natural to associate it with the cranidium. What is puzzling, however, is that this pygidium is quite different from that attributed to H. xiushanensis by Ji (1986, pi. 6, figs 6-7). This has three pairs of subequal spines, and rather subdued ring furrows. Given the depth of the glabellar furrows, an effaced axis does seem ill-suited for this species (and is unlike that of other Hadromeros species or Cheirurinae), and I think it likely that Ji’s assignment is incorrect. Genus parisoceraurus Zhou, in Wang and Jin, 1977 Type species. Parisoceraurus rectangularis Zhou, 1977, Huangnehkan Formation, Ashgill, Jianxi, China, by original designation. Parisoceraurus rectangularis Zhou, in Wang and Jin, 1977 Plate 8, figure 12 1977 Parisoceraurus rectangularis Zhou, in Wang and Jin, p. 251, pi. 75, figs 6-7. 71978 Eccoptochilel sp. indet.; Kobayashi and Hamada, pi. 2, fig. 1. Material. Cranidia, It 25430, 25471. Stratigraphical range. Lower part of Pa Kae Formation section, 0-6-14-4 m above base. Remarks. This species was described briefly from a cranidium. It is immediately distinguishable from that of Hadromeros xiushanensis in having short (tr.) glabellar furrows. Specimens from Thailand show similarly short furrows, and the palpebral lobe in anterior position opposite S3. The fixed cheeks are a little narrower than in the Chinese specimen. Forwardly placed palpebral lobes were considered a character that distinguished Parisoceraurus from Hadromeros (Tripp et al. 1989, p. 60), but as discussed above the generic placement of xiushanensis has varied. However, the cranidium discussed here is like that of P. rectangularis and does not resemble xiushanensis. Parisoceraurus zhejiangensis Ju (in Qiu et al., 1983) also has long (tr.) glabellar furrows. A fragmentary cranidium from Langkawi figured by Kobayashi and Hamada (1978, pi. 2, fig. 1) shows a similarly forward eye position and may belong to this species. Subfamily sphaerexochinae Opik, 1937 Genus sphaerexochus Beyrich, 1845 Sphaerexochus fibrisulcatus Lu, 1975 Plate 10, figures 2-3, 5 1975 Sphaerexochus fibrisulcatus Lu, pp. 218, 427, pi. 43, figs 1-4. 1977 Sphaerexochus fibrisulcatus Lu; Wang and Jin, p. 251, pi. 75, fig. 4. Material. Cranidia, It 25213-25214, 25329-25331, 25473, 25494-25495. Stratigraphical range. Pa Kae Formation, throughout measured section. Remarks. This species was described originally by Lu (1975) from the upper part of the Pagoda Limestone from the Ichang district, Hubei. He thoroughly discussed differences from previous described species. Well-preserved cranidia from Thailand clearly show the diagnostic character: effacement of the lateral glabellar furrows. However, the occipital ring is well defined. SI is visible as a faint, backward-curving line. This is unusual in sphaerexochines (indeed, all cheirurids), in which SI is strongly incised, and I know of no other species with this feature. The surface sculpture comprises fine-scale tubercles. FORTEY: ORDOVICIAN TRILOBITES 443 Family lichidae Hawle and Corda, 1847 Genus lichas Dalman, 1827 Types species. Lichas laciniatus (Wahlenberg), Dalmanites Shale (Ashgiil), Sweden, by original designation, Lichas ? sp. Plate 10, figure 12 Material. Hypostomes, It 25347, 25443. Stratigraphical range. Pa Kae Formation section, 1 4-4 — 39-0 m above base. Remarks. The only evidence for lichids is two well-preserved hypostomes, one of which is illustrated, this being the first proof of the family in the Ordovician of the Shan Thai block. It is of lichine form, and is tentatively assigned to Lichas on the basis of its similarity to the hypostome of Lichas affinis (Angelin) (e.g. Hammann 1992, pi. 29, fig. 5). Family odontopleuridae Burmeister, 1843 Genus miraspis Richter and Richter, 1917 Type species. Odontoplewa mira Barrande, 1846, Liter) Formation (Wenlock), Bohemia, by original designation. Miraspis sp. indet. Plate 10, figure 1 Material. Cranidia, It 25888-25889. Stratigraphical range. Pa Kae Formation, 18 m from the base of section. Remarks. The odontopleurid material is incomplete, but is worth recording as the only example of the family from the Ordovician of the Shan Thai block. Ramskold (1991) has revised many of the generic concepts in the family and listed species assignments. The figured small cranidium with long, paired occipital spines typical of Miraspis differs from the type species only in its wider median glabellar lobe, and very coarse surface sculpture (see, for example, Bruton 1966, pi. 7). Most other Miraspis species listed by Ramskold are Silurian, but M. solhergensis Bruton, 1966, and M. ceryx Whittington and Bohlin, 1958 are Ashgiil and Llanvirn respectively. The former has subdued sculpture and short occipital spines; the latter is much wider (tr.) than the Thai species. Both these species were erected on the basis of sparse material, but I am reluctant to add another by formally recognizing the Thai form from two cranidia. Genus hispaniaspis Hammann, 1992 Types species. Hispaniaspis dereimsi Hammann, Ordovician (Ashgiil), Cystoid Limestone, Spain, by original designation. Hispaniaspis ? sp. indet Plate 10, figures 4, 6 Material. Cranidium, It 25892; possibly associated free cheek. It 25886. Stratigraphical range. Pa Kae Formation section, 1 4-4 — 18 0 m above base. 444 PALAEONTOLOGY, VOLUME 40 Remarks. A well-preserved, small cranidium shows an inflated median glabellar lobe, subdued, but apparently fused lateral glabellar lobes, a large palpebral lobe placed in a posterior position, and apparently the circular base of a single, large median occipital spine. It seems unlikely that this spine forked distally in the manner of Dicranurus because such spines are produced by an extension of the entire occipital area, whereas the spine in the Thai specimen is discrete. A sculpture of scattered tubercles is associated with the median glabellar lobe and the fixed cheeks. I have very tentatively associated a free cheek from the same bed showing a similar sculpture beneath the eye, which is also appropriately large. L. Ramskold (pers. comm. 1995) has indicated to me that the combination of cranidial characters is unusual and that the Thai form may represent a new genus. The material at hand is not adequate to name it. The inflated median glabellar lobe can be compared to that of Whittingtonia (e.g. Hammann 1992, pi, 31, fig. 3), but the fusion and effacement of the lateral lobes is different, as is the single occipital spine and large size of the palpebral lobes and posterior position of the eye. Hispaniaspis is known only from the Spanish type species, and it, too, has paired occipital spines, although the median ‘tubercle’, with which the occipital spine is presumably homologous, is also prominent. H. dereimsi does have the eye in a similar position to the Thai species, but its lateral glabellar furrows are not fused. A free cheek (Hammann 1992, pi. 32, fig. 4) is generally like the one from Thailand, having a wide and flat exterior border, although the eye is evidently considerably smaller. I have opted to refer the Thai species with question to Hispaniaspis, under open nomenclature. This interesting species would clearly repay further research. Acknowledgements . 1 am indebted to Thanis Wongwanich and Clive Burrett for inviting me to study this fauna, and for guiding me in the field. Dr S. Bunopas approved field support from the Geological Survey of Thailand, without which the work would have been impossible. On a subsequent occasion (1994) Art Boucot supported my visit to the locality, and I express my debt both to him and to the National Science Foundation. For taxonomic advice I thank Phil Lane, Bob Owens, Lars Ramskold and Zhou Zhi-qiang. Caryl Plewes ably assisted with the preparation of specimens, and Claire Mellish with the diagrams. REFERENCES ahlberg, p. 1989. Agnostid trilobites from the Upper Ordovician of Sweden and Bornholm, Denmark. Bulletin of the Geological Society of Denmark, 37, 213-226. angelin, N. p. 1854. Palaeontologia Scandinavica, Part 1. Crustacea Formationis Transition is. Appendix. Lipsiae, Lund, 21-92, pis 25-41. apollonov, m. k. 1974. [Ashgill trilobites of Kazakhstan] Nauka Kazakhskoij SSR, Alma Ata, 136 pp., 21 pis. [In Russian], baldis, b. and pothe de baldis, E. d. 1995. Trilobites Ordovicicos de la Formacion Las Aguafditas (San Juan, Argentina) y consideraciones estratigraficas. Boletin de la Academia Nacional de Ciencias, Cordoba, 60, 409-435, pis 1-7. Bancroft, b. b. 1949. Upper Ordovician trilobites of zonal value in southeast Shropshire. Proceedings of the Roved Society of London, Series B. 136, 291-315, pis 9-11. barrande, j. 1846. Notice preliminaire sur le systeme silurien et les trilobites de Boheme. Bellman, Leipzig. vi + 91 pp. - 1852. Systeme silurien de la Boheme. lere partie. Recherches Paleontologiques. C. W. Medau, Prague and M. M. Gottlieb Haase fils, Paris, 935 pp., 541 pis. beyrich, E. 1845. tjber einiger bohmisclien Trilobiten. Berlin, 47 pp., 1 pi. billings, e. 1865. Palaeozoic fossils, vol. 1. Geological Survey of Canada. Ottawa, 426 pp. bruton, d. l. 1966. A revision of the Swedish Ordovician Odontopleuridae (Trilobita). Bulletin of the Geological Institutions, University of Uppsala , 43 (8), 1-40, pis 1-6. 1968. The trilobite genus Panderia from the Ordovician of Scandinavia and the Baltic areas. Norsk Geologisk Tidsskrift, 48, 1-34, pis 1-12. 1976. The trilobite Phillipsinella in the Ordovician of Great Britain and Scandinavia. Palaeontology, 19, 699-714, pis 104-108. and owen, a. w. 1988. The Norwegian Upper Ordovician illaenid trilobites. Norsk Geologisk Tidsskrift, 68, 241-258. FORTEY: ORDOVICIAN TRILOBITES 445 burmeister, H. 1843. Die organisation der Trilobiten aus ihren lebenden Verwandten entwickelt; nebst einer systematischen Uebsersicht alter zeither beschriebenen Arten. Berlin, xx+ 147 pp., 6 pis. burrett. c. F., long, J. and stait, b. 1990. Early-Middle Palaeozoic biogeography of Asian terranes derived from Gondwana. 163-174, In McKERROW, w. s. and SCOTESE, c. r. (eds). Palaeozoic palaeogeography and biogeography. Memoirs of the Geological Society of London, 13, 435 pp. BURST, J. F. 1965. Subaqueously formed shrinkage cracks in clay. Journal of Sedimentary Petrology , 35, 348-353. CHEN XU, RONG JIA-YU, WANG XIAO-FENG, WANG ZHI-HAO, ZHANG YUAN-DONG and ZHAN REN-BIN 1995. Correlation of the Ordovician rocks of China. International Union of Geological Sciences Publication , 31, 1-104. cocks, l. r. m. and fortey, r. a. 1988. Lower Palaeozoic facies and faunas around Gondwana, 183-200. In hallam, a. and audley-charles, m. (eds). Gondwana and Tethys. Special Publications of the Geological Society, London, 37, 317 pp. — and rong jia-yu 1988. A review of the later Ordovician Foliomena brachiopod fauna. Palaeontology , 31, 53-67. dalman, J. w. 1827. Om palaeaderna eller de sa kallade Trilobiterna. Kunglinga Svenska Vetenskapsaka- demiens Handlingar, 113-152, 226-294, pis 1-6. [For 1826]. DEAN, w. T. 1963. The Ordovician faunas of South Shropshire, 3. Bulletin of the British Museum (Natural History), Geology Series , 7, 213-254, pis 31-46. 1967. The distribution of Ordovician shelly faunas in the Tethyan region. I 1^14. In adams, c. g. and ager, d. v. (eds). Aspects of Tethyan biogeography. Systematics Association Publication, 7, London, 336 pp. 1971 Trilobites of the Chair of Kildare Limestone (Upper Ordovician) of eastern Ireland. Part I. Monograph of the Palaeontographical Society, 125 (531), 1-64, pis 1-25. — and zhou zhi-yi 1988. Upper Ordovician trilobites from the Zap Valley, south-east Turkey. Palaeontology, 31, 621-649. dolambi, T. and gond, p. 1991. Aocaspis nov. gen., nouveau trilobite Nileidae de l’Ordovicien Inferieur de Montagne Noire (France). Geobios , 25, 213-224. fortey, R. a. 1975u. Early Ordovician trilobite communities. Fossils and Strata, 4, 331-352. — 19756. The Ordovician trilobites of Spitsbergen. 2. Asaphidae, Nileidae, Raphiophoridae and Telephinidae of the Valhallfonna Formation. Skrifter Norsk Polarinstitutt , 162, 1-125, pis 1-41. — 1980. The Ordovician trilobites of Spitsbergen. 3. Remaining trilobites of the Valhallfonna Formation. Skrifter Norsk Polarinstitutt, 171, 1-113, pis 1-25. — 1985. Pelagic trilobites as an example of deducing the life habits of extinct arthropods. Transactions of the Royal Society of Edinburgh: Earth Sciences, 76, 219-231. 1990. Ontogeny, hypostome attachment and trilobite classification. Palaeontology, 33, 529-576. — and clarkson, E. N. K. 1976. The function of the glabellar ‘tubercle' in Nileus and other trilobites. Lethaia, 9, 101-106. — harper, d. a. T., ingham, J. K., owen, a. w. and rushton, a. w. a. 1995. A revision of Ordovician series and stages from the historical type area. Geological Magazine, 132, 15-30. — and owens, R. M. 1987. The Arenig Series in South Wales. Bulletin of the British Museum (Natural History ), Geology Series, 41, 69-307. — and wilmot, n. 1991. Trilobite cuticle thickness in relation to palaeoenvironment. Paldontologische Zeitschrift, 65, 141-151. fu li-pu hu, yun-xu, zhang zi-fu and wang shu-xi 1993. The mark on the ecology of the sedimentary environment in Middle and Upper Ordovician at Ordos Basin. Northwest Geoscience, 14, 1-88, pis 1-8 [In Chinese, English summary]. hammann, w. 1992. The Ordovician trilobites from the Iberian Chains in the province of Aragon, NE-Spain. I. The trilobites of the Cystoid Limestone (Ashgill Series). Beringeria, 6, 1-144, pis 1-37. han nai-ren 1980. [On the carapace of Hammatocnemis Kielan ] Geological Review, Beijing, 26, 183-189, pis 1-2 [In Chinese]. hawle, i. and corda, a. j. c. 1 847. Prodrom einer monographic der bohmischen trilobiten. Prague, 176 pp., 7 pis. henry, j.-l. 1994. The trilobites of the ‘affaire Deprat’. Alcheringa, 18, 359-362. horny, r. and bastl, f. 1970. Type specimens of fossils in the National Museum, Prague, Volume 1, Trilobita. National Museum, Prague. 354 pp., 20 pis., howell, b. f. 1935. Cambrian and Ordovician trilobites from Herault, southern France. Journal of Paleontology, 9. 222-238, pi. 23. 446 PALAEONTOLOGY, VOLUME 40 hu chung-hung 1971. Ontogeny and sexual dimorphism of Lower Paleozoic Trilobita. Palaeontographica Americana , 44, 1-130, pis 1-26. hunt, a. s. 1967. Growth, variation and instar development of an agnostid trilobite. Journal of Paleontology, 41, 203-208. JAEKEL, o. Uber die Agnostiden. Zeitschrift der Deulschen Geologischen Gesellschaft , 61, 380-401. ji zai-liang 1982. [Stratigraphical distribution and range of the trilobite Family Phillipsinellidae in South China.] Bulletin of the Institute of Geology, Chinese Academy of Geological Sciences, 6, 57-68, pis 1-2. [In Chinese]. 1985. [On the depositional environment of the Pagoda Formation in central and southwestern China.] Professional Papers in Palaeontology and Stratigraphy, 12, 87-93, pis 1-2. [In Chinese]. 1986. [Upper Ordovician (middle Caradoc-early Ashgill) trilobites from the Pagoda Formation in South China.] Professional Papers in Palaeontology and Stratigraphy, 15, 1-33, pis 1-6. [In Chinese], kielan, z. 1957. On the trilobite family Staurocephalidae. Acta Palaeontologica Polonica, 2, 155-180, pis 1-6. 1960. Upper Ordovician trilobites from Poland and some related forms from Bohemia and Scandinavia. Palaeontologia Polonica, 11, 1-198, pis 1-36. kobayashi, t. 1939. On the agnostids. Journal of the Faculty of Science, Tokyo University, Section 2, 5, 66-198. - and hamada, t. 1970. Contributions to the geology and palaeontology of SE Asia. 78. A cyclopygid- bearing Ordovician faunule discovered in Malaysia with a note on the Cyclopygidae. Geology and Palaeontology of Southeast Asia, 8, 1-18, pis 1-2. 1978. Upper Ordovician trilobites from the Langkawi Islands, Malaysia. Geology and Palaeontology of Southeast Asia , 19, 1-28, pis 1-2. koroleva, M. N. 1959a. [New genera of trilobites from the Middle and Upper Ordovician of northern Kazakhstan.] Doklady Akademia Nauk SSSR , 124, 1313-1316. [In Russian]. - 1959A [New trilobites from the upper Ordovician (Caradoc) of Northern Kazakhstan ] Sbornik Kazakhskogo gorno-metallurgicheskogo Instituta, 20, 75-1 13, pis 1-4. [In Russian], - 1982. [ Ordovician trilobites of north-east Kazakhstan.] Ministry of geology of the Kazakh SSR (KazIMS). Moscow, 152 pp, 26 pis. [In Russian], land, M. f. 1981. Optical mechanisms in the higher Crustacea with a comment on their evolutionary origins, 31-48. In laverack, M. s. and cosens, d. s. (eds). Sense organs. Blackie, Glasgow, 289 pp. 1984. Crustacea. 401-438. In ali, m. a. (ed.). Photoreception and vision in invertebrates. Plenum, New York, 853 pp. lane, p.d. 1971. British Cheiruridae (Trilobita). Monograph of the Palaeontographica I Society , 125 (530), 1-95, pis 1-16. leone, f., hammann, w., laske, r. serpagli, e. and villas, e. 1991. Lithostratigraphical units and biostratigraphy of the post-Sardic Ordovician sequence in southwestern Sardinia. Bolletina della Societa Paleontologica Italiana, 30, 201-235. li yao-xi, song li-sheng, zhou zhi-qiang and yang jing-yao 1975. [Early Palaeozoic stratigraphy of the western section of Tapa Mountain.] Earth Sciences Publishing House, Beijing, 232 pp., 70 pis. [In Chinese]. linnarsson, j. g. o. 1869. Om Vastergotlands Cambriska och Siluriska aflagringar. Kongliga Svenska Vetenskaps-Akadiemens Handlingar, 8, 1-89, pis 1-2. lu yan-hao 1957. Trilobita, 249-294, pis 137-155. In [Index fossils of China , Invertebrate.] Institute of Palaeontology Academica Sinica. [In Chinese]. 1962. [Index fossils of the Yangtze Region.] Science Press, Beijing. [In Chinese], 1964. [Handbook of Index fossils of Southeast China] Science Press, Beijing. [In Chinese], 1 975. Ordovician trilobite faunas of central and southwestern China. Palaeontologica Sinica, New Series B, 11, 1-463, pis 1-50. — and chang wen-tang 1974. [Ordovician trilobites.] In [A handbook of stratigraphy and palaeontology in southwest China] Science Press, Beijing. [In Chinese]. — CHU CHAO-LING, CHIEN YI-YUAN, ZHOU ZHI-YI, CHEN JUN-YUAN, LIU GENG-WU, YU WEN, CHEN XU and XU han-kui 1976, Ordovician biostratigraphy and palaeozoogeography of China. Memoirs of the Nanjing Institute of Geology and Palaeontology, 7, 1-83, pis 1-14. [In Chinese, English summary]. and zhou zhi-yi 1979. [Systematic position and phylogeny of Hammatocnemis (Trilobita).] Acta Palaeontologica Sinica, 18, 415-433, pis 1-4. [In Chinese]. 1981. [Early Upper Ordovician trilobites from the Nanjing Hills], Bulletin of the Nanjing Institute of Geology and Palaeontology, 3, 1-27, pis 1-7. [In Chinese]. marek, l. 1952. Contribution to the stratigraphy and faunas of the upper part of the Kraluv Dvur Shales (Ashgillian). Sbornik Ustredm'ho Ustavu Geologickeho, 19, 429-455. FORTEY: ORDOVICIAN TRILOBITES 447 — 1960. The trilobite family Cyclopygidae Raymond in the Ordovician of Bohemia. Rozpravy Ustredniho Ustavu Geologickeho , 28, 1-85, pis 1-6. M COY, f. 1845. A synopsis of the Silurian fossils of Ireland. Sir R. J. Griffith, Dublin, 72 pp., 5 pis. metcalfe, i. 1992. Ordovician to Permian evolution of Southeast Asian terranes: NW Australian Gondwana connections, 293-306. In WEBBY, b. d. e. and laurie, j. r. (eds). Global perspectives on Ordovician geology. A. A. Balkema, Rotterdam, 513 pp. miller, s. a. 1889. North American geology and palaeontology. Mattocks, Cincinnati, 664 pp. moore, r. c. (ed.). 1959. Treatise on invertebrate paleontology. Part O. Arthropoda I. University of Kansas Press and Geological Society of America, Lawrence, Kansas, 560 pp. morris, s. F. 1988. A review of British trilobites, including a synoptic revision of Salter’s monograph. Monograph of the Palae onto graphical Society, 140 (574), 1-316. nielsen, a. t. 1995. Trilobite systematics, biostratigraphy and palaeoecology of the Lower Ordovician Komstad Limestone and Huk Formations, southern Scandinavia. Fossils and Strata , 38, 1-374. nibcolaisen, F. 1983. The Middle Ordovician of the Oslo Region, Norway, 32. Trilobites of the family Remopleurididae. Norsk Geologisk Tidsskrift , 62, 231-299, pis 1-15. opik, a. a. 1937. Trilobiten aus Estland. Acta et Commentationes Universitatis Tartuensis , 32, 1-137, pis 1-26. owen, a. w. and bruton, d, l. 1980. Late Caradoc-early Ashgill trilobites of the central Oslo Region, Norway. Palaeontological Contributions of the University of Oslo, 245, 1-42, pis 1-10. — bruton, d. l., bockelie, j. f. and bockelie, T. G. 1990. The Ordovician successions of the Oslo Region, Norway. Special Publication of the Norwegian Geological Survey, 4, 1-54. owens, R. M. 1973a. Ordovician Proetidae (Trilobita) from Scandinavia. Norsk Geologisk Tidsskrift, 53, 117-181. — 1973A British Ordovician and Silurian Proetidae (Trilobita). Monograph of the Palaeontographical Society , 127 (535), 1-98, pis 1-15. 1979. The trilobite genera Panarchaeogonus Opik, Isbergia Warburg and Cyamops gen. nov. from the Ordovician of Balto-Scandia and the British Isles. Norsk Geologisk Tidsskrift, 58, 199-219. and hammann, w. 1990. Proetide trilobites from the Cystoid Limestone (Ashgill) of NW Spain, and the suprageneric classification of related forms. Palaontologische Zeitschrift, 64, 221-244. pek, i. 1977. Agnostid trilobites of the central Bohemian Ordovician. Sbornik Geologickych Ved, 1977 (1), 7-44, pis 1-12. — and prokop, r. j. 1984. New finds of the agnostid trilobites from the Ordovician of the Prague area (Czechoslovakia). Casopis Narodniho Muzea, 153, 17-20, pi. 1. portlock, J. F. 1843. Report on the geology of the Country of Londonderry , and of parts of Tyrone and Fermanagh. Geological Survey of Ireland, Dublin and London, xxi + 784 pp., 54 pis. prantl, F. and pribyl, a. 1948. Classification of some Bohemian Cheiruridae. Sbornik Narodniho Muzea v Praze, 3B. 1-44. price, d. and magor, p. M. 1984. The ecological significance of variation in the generic composition of Rawtheyan (late Ordovician) trilobite faunas from North Wales, UK. Geological Journal, 19, 187-200. qiu, hong-an and 10 others. 1983. [Paleontological Atlas of eastern China. 1. Lower Palaeozoic .] Geological Press, Beijing, 657 pp., 176 pis. [In Chinese], ramskold, l. 1991. Pattern and process in the evolution of the Odontopleuridae (Trilobita). The Selenopeltinae and Ceratocephalinae. Transactions of the Royal Society of Edinburgh: Earth Sciences, 82, 143-181. Raymond, p. e. 1925. Some trilobites of the lower Middle Ordovician of eastern North America. Bulletin of the Museum of Comparative Zoology, Harvard University, 47, 1-180. reed, f. R. c. 1896. The fauna of the Keisley Limestone. Part 1. Quarterly Journal of the Geological Society, London, 52, 401-437. — 1935. The Lower Palaeozoic trilobites of Girvan. Supplement No. 3. Monograph of the Palaeontographical Society, 88 (400), 1-64, pis 1-4. richter, r. and richter, e. 1917. Uber die Einteilung der Familie Acidaspidae und fiber einige ihrer devonischen Vertreter. Zentralblatt fiir Mineralogie, Geologie und Paldontologie, 1917, 462-472. salter, J. W. 1864. A monograph of the British trilobites from the Cambrian, Silurian and Devonian formations. Part 1. Monograph of the Palaeontographical Society, 16 (67), 1-80, pis 1-6. sars, G. o. 1835. Uber einige neue oder unvollstandig bekannte trilobiten. Isis, 4, 333 — 43, pi. 9. schrank, e. 1972. Nileus-Arten (Trilobita) aus Geschieben des Tremadoc bis tieferen Caradoc. Berichte der Deutschen Gesellschaft fiir Geologische Wissenschaften, Berlin, 17, 351-375. sheng shen-fu. 1964. Upper Ordovician trilobite faunas of Szechuan-Kweichow with special discussion on the classification and boundaries of the Upper Ordovician. Acta Pa/aeontologica Sinica, 12, 553-581. 448 PALAEONTOLOGY, VOLUME 40 sheng xin-fu and n zai-liang 1987. [On the age of the Pagoda Formation ] Professional Papers in Palaeontology and Stratigraphy , 16. 1-32, pis 1-2. [In Chinese]. shergold, j. s., laurie, j. r. and sun, x.-w. 1990. Classification and review of the trilobite order Agnostida Salter, 1864: an Australian perspective. Report of the Bureau of Mineral Resources , Geology and Geophysics , Australia , 296, 1-93. snajdr, m. 1956. The trilobites from the Drabov and Letna beds of the Ordovician of Bohemia. Sbornik Ustredniho Ustavu Geologickeho, 22, 1-57. stait, b. and BURRF.TT, c. f. 1984. Ordovician nautiloid faunas of central and southern Thailand. Geological Magazine , 121, 115-124. and wongwanich, t. 1984. Ordovician trilobites from the Tarutao Formation, southern Thailand. Neues Jahrbuch fur Mineralogie , Geologie und Paldontologie, Monatshefte , 1984, 53-64. strauss, d. and sadler, p. m. 1989. Classical confidence intervals and Bayesian probability estimates for ends of local taxon ranges. Mathematical Geology , 21, 41 1 — 427. temple, j. t. 1965. The trilobite genus Oedicybele from the Kildare Limestone (Upper Ordovician) of Eire. Palaeontology , 8, 1-4, pi. 1. Tripp, r. p. 1962. Trilobites of the ‘Confinis’ Flags (Ordovician) of the Girvan District, Ayrshire. Transactions of the Royal Society of Edinburgh , 65, 1 40, pis 1-4. zhou zhi-yi and pan zhen-qin 1989. Trilobites from the Upper Ordovician Tangtou Formation, Jiangsu Province, China. Transactions of the Royal Society of Edinburgh: Earth Sciences. 80. 25-68. ulrich, e. o. 1930. Ordovician trilobites of the family Telephidae and concerned stratigraphical correlations. Proceedings of the US National Museum , 76, 1 101, pis 1-8. vogt, k. 1980. Die Speigeloptik des Flusskrebsauges. The optical system of the crayfish eye. Journal of Comparative Physiology , 135, 1-19. volborth, a. von 1863. Uber die mit glatten Rumpfgliedern versehenen russischen Trilobiten, nebst einem Anhange fiber die Bewegungsorgane und fiber das Herz derselben. Memoires de I'Academie Imperiale des Sciences, St Petersbourg, Series 7, 6 (2), 1-47, pis 1-4. wang, xaio-feng and jin yu-QIN (eds). 1977. [ Palaeontology of the Central-Southern region of China (Vol. 1).] Geological Press, Beijing, 470 pp., 1 16 pis. [In Chinese]. warburg, e. 1925. The trilobites of the Leptaena Limestone in Dalarne. Bulletin of the Geological Institutions of the University of Uppsala, 17, 1 — 446, pis 1-11. weber, v. n. 1948. [Trilobites of the Silurian sediments of the USSR: Lower Silurian trilobites.] Monographs of the Palaeontology of the USSR, Ministry of Geology, Moscow, 113 pp., 11 pis. [In Russian], weir, J. a. 1959, Ashgill trilobites from Co. Clare Ireland. Palaeontology , 1, 369-383, pis 62-63. whittard, w. f. 1966. The Ordovician trilobites of the Shelve Inlier, South Shropshire. Part 8. Monograph of the Palaeontographical Society, 119 (504), 265-306, pis 46-50. Whittington, h. b. 1938. The geology of the district around Llansantffraid ym Mechain, Montgomeryshire. Quarterly Journal of the Geological Society, London , 94, 423-60, pis 38-39. 1950. Sixteen Ordovician genotype trilobites. Journal of Paleontology, 24, 531-565, pis 68-75. 1963. Middle Ordovician trilobites from Lower Head, western Newfoundland. Bulletin of the Museum of Comparative Zoology , Harvard University, 129, 1-118, pis 1-36. 1965. Trilobites of the Ordovician Table Head Formation, western Newfoundland. Bulletin of the Museum of Comparative Zoology , Harvard University, 132, 27 5 — 442, pis 1-68. — and bohlin, b. 1958. New Ordovician Odontopleuridae (Trilobita) from Oland. Bulletin of the Geological Institutions, University of Uppsala, 38, 37-45. wilmot, n. 1990. Cuticular structure of the agnostine trilobite. Homagnostus obesus. Lethaia, 23, 87-92. wongwanich, t., burrett, c. f., tansathien, w. and chaodumrong, p. 1990. Lower to mid-Palaeozoic stratigraphy of mainland Satun Province, southern Thailand. Journal of Southeast Asian Earth Sciences, 4, 1-9. XIA SHU-FANG. 1978. Ordovician Trilobita. 167-185, pis 28-37. In research group of stratigraphy of YANGTZE GORGES, geological bureau OF hubei province (eds). [Sinian to Permian stratigraphy and palaeontology of the eastern Yangtze Gorges .] Geological Publishing House, Beijing, 43 pp., 67 pis. [In Chinese]. xiang Li-WF.N and ji zai-liang 1986. [Upper Ordovician (Ashgill) trilobites from the Linxiang Formation of western Hunan and eastern Guizhou.] Bulletin of the Chinese Academy of Geological Sciences, 12. 53-66, pis 1-2. [In Chinese]. zhang wen-tan 1979. On the Miomera and Polymera (Trilobita). Scientia Sinica, 10. 996-1004, 1 pi. [In Chinese, reprinted in English in 1980, Scientia Sinica, 23, 223-234], FORTEY: ORDOVICIAN TRILOBITES 449 ZHENG QING-LUAN, NI SHI-ZHAO, XU GUANG-HONG, ZHOU TIAN-MEI, WANG XIAO-FENG, LI ZHI-HONG, LAI CAI-GEN and xiang, Li- wen 1983 [Subdivision and correlation of the Ordovician in the eastern Yangtze Gorges.] Bulletin of the Yichang Institute of Geology and Mineral Resources , Chinese Academy of Geological Sciences , 10, 1-56. [In Chinese]. zhou zhi-qiang (with lee, j.-s. and qu, xin-gou). 1982. Trilobita. 215^460. In xi'an institute of geological and mineralogical research (eds). [ Palaeontological atlas of northwest China: Shaanxi, Gansu and Ningxia. Volume I. Precambrian to early Palaeozoic .] Geological Publishing House, Beijing, 1072 pp. [In Chinese]. — and xiang li-wen 1993. Proetida (Trilobita) from the Pagoda Limestone (Caradoc) of northern Upper Yangtze Platform, China. Stratigraphy and Paleontology of China , 2, 51-75, pis 1-4. zhou zhi-yi 1987. Notes on Chinese Ordovician agnostids. Acta Palaeontologica Sinica , 26, 639-661. and dean, w. t. 1986. Ordovician trilobites from Chedao, Gansu Province, north-west China. Palaeontology , 29, 743-786. - 1989. Trilobite evidence for Gondwanaland in east Asia during the Ordovician. Journal of Southeast Asian Earth Science, 3, 131-140. Typescript received 17 December 1995 Revised typescript received 25 June 1996 RICHARD A. FORTEY The Natural History Museum Cromwell Road London SW7 5BD, UK NOTE ADDED IN PROOF Since this paper was accepted 1 have seen the type species of Brontocephalina Chugaeva, B. marginatula , from the upper Ordovician of north-east Russia (chugaeva, m. n. 1975. [Late Ordovician trilobites of the north-east of the USSR.] Transactions of the Geological Institute, Academy of Sciences of the USSR , 272, 1-63, pis 1-1 1 [In Russian]) which shows similar pygidial structure to ‘ Cekovia ’ transversa. The Thai species is better referred to Brontocephalina than to Cekovia. BUBBLE-HEADED TRILOBITES, AND A NEW OLENID EXAMPLE by r. a. fortey and r. m. owens Abstract. Several trilobites developed an inflated cephalic lobe with a distinctive bubble-like profile. This happened polyphyletically in at least seven families ranging from the Cambrian to the Silurian. We describe a new species of Ordovician (Tremadoc) olenid trilobite, Parabolinella bolbifrons, having this morphology. A review of other trilobites with apparently similar cephala shows that the bubble-headed appearance was derived in several different ways and probably acquired different functions. In deiphonine cheirurids, staurocephalids and Paraphillipsinella all, or only the frontal part of the glabella, is involved in the inflation. These species have rigidly attached conterminant hypostomes, and glabellar inflation may have matched a comparable inflation of the stomach. In the new species, as in several Cambrian examples with natant hypostomes, inflation is confined to a median area of the preglabellar field. This is unlikely to have involved any modification of the alimentary system. Trilobites displayed many modifications to their dorsal shields through their 300 million year history. The cephalon was particularly prone to drastic evolutionary change, such as loss of the eyes, general eflfacement, or becoming covered in coarse tuberculation. When we were reviewing evolutionary trends in trilobites (Fortey and Owens 1990) we recognized certain morphologies that had arisen repeatedly during trilobite history. Usually, these were morphological packages involving the entire exoskeleton. We termed these recurring designs morphotypes. The assumption was made that such morphotypes represented a specific mode of life, which trilobites adopted repeatedly. The olenid morphotype, for example, included such features as a low cephalic and axial convexity, multiplication of thoracic segments, which became narrow (sag.), but with wide pleurae, and a notably thin exoskeleton. The olenid morphotype arose on several occasions from different phylogenetic origins; we regarded it as an adaptation for life in oxygen-poor (dysaerobic) habitats. Another morphological modification that impressed us concerned the cephalon. We found a number of trilobites in which some part of the axial region was inflated dramatically into a perfectly spherical balloon, or bubble-like structure. These are the bubble-headed forms that we discuss in this paper, and we discuss whether or not these remarkable trilobites comprise another morphotype. During the same period we were also investigating the faunas of the British Tremadoc (Fortey and Owens 1 99 1 Z?, 1992). In the course of this work we revised the olenid trilobite Beltella (Fortey and Owens 1991r/) from material in the Bristol City Museum collected by M. L. K. Curtis and T. R. Fry from temporary exposures of the Breadstone Shales in the Tortworth Inlier, Gloucestershire. Unlike many Tremadoc fossils in the British Isles this material has suffered little tectonic distortion. In the same collection there is another, remarkable olenid assigned to the genus Parabolinella , which provides a beautiful example of bubble-head morphology; this new species is described below. PHYLOGENETIC ORIGINS OF BUBBLE-HEADED TRILOBITES It is clear that the bubble-headed morphology may have been derived from more than one phylogenetic source. It is a morphology which has been described from several families ranging from the Cambrian to the Silurian. In a few groups (Staurocephalidae, Nepeidae) the bubble-head IPalaeontology, Vol. 40, Part 2, 1997, pp. 451—459, 1 pl.| © The Palaeontological Association 452 PALAEONTOLOGY, VOLUME 40 construction has been accorded familial significance. In other cases, one or more genera within a family develop this feature, while its close relatives are more ‘normal’ trilobites. Many more trilobites exhibit a certain convexity in the antero-median region, without, however, displaying the independent convexity associated with bubble head morphology. In any case, there is no suggestion that all bubble-heads should be classified together. A selection of these trilobites is illustrated in Text-figure 1 . The following families include examples: Olenida e: - Parabolinella bolbifrons , described below; Phillipsinellidae - Paraphillipsinella globosa (Lu in Lu and Chang, 1974) is a typical example from the upper Ordovician of China and Thailand; Cheiruridae - Deiphon barrandei Whittard, 1934 (Silurian); Sphaerocoryphe kingi Ingham, 1974 (upper Ordovician); Onycopyge liversidgei Woodward, 1880 (Upper Silurian); Staurocephalidae - Staurocephcilus susanae Thomas, 1981 (Silurian); Nepeidae - Nepea Whitehouse, 1939 (Middle Cambrian) and allied genera; Concory- phidae - a typical example is Ctenocephalus barrcmdei Hawle and Corda, 1847 (Middle Cambrian; see Snajdr 1990, pi. 100). There are several additional genera whose familial affinities are under debate. The genus Amzasskiella Poletaeva, 1960 is a late Cambrian trilobite widespread through central and eastern Asia. Toxotis Wallerius, 1895 (middle Cambrian, Sweden) is probably not related to Nepea. Jell (1985) described another example, Natmus , from the lower Ordovician (Tremadoc) of Tasmania. Tumid glabellas which approach ‘bubble-head’ morphology in their convexity are known from such families as the Trinucleidae (e.g. Tretaspis ceriodes (Angelin, 1854) -see Owen 1980), Eodiscidae (e.g. Acimetopus bilobatus Rasetti, 1966) and Aulacopleuridae (e.g. Cyphaspis hydrocephala Roemer, 1843; see Pribyl and Vanek 1981). In the classification system employed in the Treatise (Moore 1959) or modified by Fortey (1990), Cheiruridae and Staurocephalidae are placed in the order Phacopida; Phillipsinellidae was assigned by Bruton (1976) and Fortey (1990) to the suborder Scutelluina, and possibly the order Corynexochida. Olenidae, Nepeidae and Conocoryphidae could be classified among the Ptychopariida. Hence the morphology arose independently in three different orders, and more if the examples with tumid glabellas are included. Furthermore, there is evidence that even within a given order the modification occurred independently in each example. Staurocephalus probably does not share a common ancestor with Deiphon and Sphaerocoryphe (Fane 1971). The olenid described below is evidently closer to other species in the same genus, Parabolinella, than it is to Nepea or Ctenocephalus. Paraphillipsinella is related to Phillipsinella, a small trilobite with a tumid glabella but hardly a spherical one. We consider it likely therefore, that bubble-headedness arose independently in all these examples. This carries with it the implication that for all its distinctiveness this drastic cephalic modification enjoys a relatively trivial phylogenetic burden. It could be relatively easily acquired. In this case the next question to address is whether the feature is homologous or even analogous in the different examples cited. In other words, it can be questioned whether bubble-heads truly characterize a morphotype, a design likely to have shared a common life habit. ARE BUBBLE-HEADED TRILOBITES A DISTINCT MORPHOTYPE? We consider it improbable that bubble-head development is homologous, or even analogous, in the various cases in which we have observed it. This is because the inflation in various examples is manifested in different parts of the cephalic axial anatomy. Deiphon type (Text-fig. If, h). Much of glabella inflated in front of the occipital ring In Deiphon (see Lane 1971; Holloway 1980, p. 39) and Onycopyge (see Holloway and Campbell 1974) inflation affects much of the glabella in front of the occipital segment. A relict LI remains as a pair of inconspicuous nodes within the preoccipital depression. In Sphaerocoryphe this portion of the glabella is more conspicuous. FORTEY AND OWENS: BUBBLE-HEADED TRILOBITES 453 text-fig. 1. Examples of bubble-head morphology, a, Olenidae. Parabolinella bolbifrons sp. nov., with lateral view showing inferred position of hypostoma (based on specimens illustrated in Plate 1). B, Nepeidae. Ferenepea hispida Opik, 1967, Middle Cambrian, north-west Queensland (after Opik, 1967, pi. 39, hg. 8). c, Triplacephalidae. Amzasskiella [= Triplacephalus] sanduensis (Lu and Qian in Wang and Jin, 1977), Ordovician, Tremadoc Series, Guizhou Province, China (after Lu and Qian 1977, pi. 48, fig. 4). d, Conocoryphidae. Ctenocephalus howelli Resser, 1937, Middle Cambrian, south-east Newfoundland (after Hutchinson, 1962, pi. 12, fig. 18). e, Staurocephalidae. Staurocephalus susanae Thomas, 1981, Silurian, Wenlock Series, central England, dorsal and lateral views; arrow indicates hypostomal suture (after Thomas 1981, text-fig. 8, p. 66 and pi. 17, figs 2, 5, 9). f, Cheiruridae, Deiphoninae. Sphaerocoryphe kingi Ingham, 1974, Ordovician, Ashgill Series, Northern England (after Ingham 1974, text-fig. 22, p. 72). G, Phillipsinellidae. Paraphillipsinella globosa Lu in Lu and Chang, 1974, Ordovician, Caradoc Series, Gansu Province, China (after Zhou and Dean 1986, pi. 62, fig. 13). h, Cheiruridae, Deiphoninae. Deiphon barrandei Whittard, 1934, Silurian, Wenlock Series, central England, dorsal and lateral views; arrow indicates hypostomal suture (after Lane 1971, pi. 12, figs 1. 4, 6, 9). Staurocephalus type (Text-fig. 1 E, g). Frontal lobe of glabella inflated In Paraphillipsinella and Staurocephalus inflation affects only the anterior lobe of the glabella, while the posterior part of the glabella retains normal convexity. Nepea type (Text-fig. I a-d). Preglabellar field inflated In Nepea , as in Parabolinella bolbifrons described below, it is clear that the inflated area is derived entirely from that part of the exoskeleton lying anterior to the glabella. It is the preglabellar field 454 PALAEONTOLOGY, VOLUME 40 in its median region which is transformed. A similar origin applies for the inflated areas in Ctenocephalus, Amzasskiella , Toxotis and Natmus. These three cases are all different; they are certainly not homologous. It is extremely doubtful whether they are an expression of adoption of a similar life habit. Ventrally, the structure of Parabolinella bolbifrons differs remarkably from that of cheirurid and Paraphillipsinella examples. We show below that the bulb is underlain by the preglabellar field in P. bolbifrons. Since the doublure is demonstrably narrow, and the natant hypostome is likely to have resided beneath the frontal lobe of the glabella, it is clear that the posterior part of the inflated area was underlain only by the ventral membrane (see Text-fig. 2), and it is likely that this also applied to the examples known from the Cambrian. However, in the younger examples, where all or part of the glabella is inflated, the ventral side of the bulb is continued posteriorly by the hypostome, together with the border and rostral plate (Text-fig. 1e, h; Lane 1971, pi. 13, fig. 12). Since the oesophagus and stomach were located under the cephalic axis it seems reasonable to infer that in these bubble-heads the stomach had become inflated likewise. This may reflect a change in diet, for example. This cannot have been the case in Parabolinella bolbifrons in which the inflation is unconnected with the glabella, and indeed specifically excluded it. Ruedemann (1934) and Whittard (1934) regarded Deiphon and Onycopyge as planktic, arguing that the swollen glabella may have contained a low-density substance to aid flotation. Convincing arguments against this theory were advanced by Holloway and Campbell (1974). It is likely that all bubble heads were part of the benthos. Olenids such as Parabolinella appear to have been adapted to life in environments with low oxygen (dominated by this family; Henningsmoen 1957), whereas the other bubble-heads appear to have been inhabitants of diverse, open-shelf faunas. Hence, despite a superficial similarity, bubble-headed trilobites are not comparable with regard to their structure, nor likely to have shared a common adaptation. There appear to be two major ways in which the morphology can be derived : by inflation of the preglabellar area, or by inflation of all or part of the glabella itself. These two morphologies are associated with natant trilobites, on the one hand, and conterminant ones on the other. There is no single bubble-head morphotype. This example shows that it is necessary to understand fully the genesis of a striking structure before assuming that all trilobites displaying it were the product of a similar adaptation. SYSTEMATIC PALAEONTOLOGY Family olenidae Burmeister, 1 843 Genus parabolinella Brogger, 1882 Type species. Parabolinella limitis , Brogger, Tremadoc, Norway. EXPLANATION OF PLATE 1 Figs 1-8. Parabolinella bolbifrons sp. nov. ; Ordovician, Tremadoc Series, Breadstone Shales; Tortworth Inlier, Gloucestershire. 1, BRSMG Cc2133; cranidium showing pits in border furrow; x 1-5. 2, 6, BRSMG Cb4807; incomplete dorsal exoskeleton; 2, x 1 -5; 6, latex cast from the counterpart; x 1. 3, BRSUG 26361 ; large, incomplete exoskeleton; x 1 ; 4. BRSMG Cc2132; doublure of free checks, preserved from the ventral side; x L5. 5, 7, BRSMG Cc2 139; cranidium, preserving some relief; 5, x 2; 7 latex from counterpart; x3. 8, BRSMG Cc2134; poor cranidium with well-preserved bulb; x 1.5. Figs 2, 6, holotype; all others paratypes. All internal moulds unless stated otherwise. BRSMG: Bristol City Museum; BRSUG: University of Bristol, Department of Geology. PLATE 1 FORTEY and OWENS, Parabolinella 456 PALAEONTOLOGY, VOLUME 40 text-fig. 2. Parabolinella bolbifrons sp. nov.; recon- struction of ventral side of cephalon, showing inferred position of hypostoma. Parabolinella bolbifrons sp. nov. Plate 1, figures 1-8; Text-figures 1a, 2 1996 Parabolinella sp. nov. Owens, p. 69, pi. 1, figs c-d. Derivation of name. Latin, referring to its characteristic median anterior inflation. Holotype. Incomplete dorsal exoskeleton, Bristol City Museum, BRSMG Cb4807. Paratypes Incomplete dorsal and cephalic shields BRSUG 26361 (Bristol University), BRSMG Cc2127, Cc2128, Cc2133, Cc2134, Cc2139 Cc2141 and a doublure of yoked cheeks BRSMG Cc2132. Occurrence. All material is from exposures of the Breadstone Shales (Tremadoc) in a temporary trench 192 m W 32° S of Crawless Old Barn, south-south-west of Breadstone, in the Tortworth Inlier, Gloucestershire [ST 7026 9997], Curtis (1968) described the stratigraphy of this area and has made extensive new collections from temporary exposures in the soft shales. Fortey and Owens (1991a) redescribed the olenid trilobite Beltella depressa from the same exposures; this species occurs elsewhere in the Lower Tremadoc (Cressagian Stage in the terminology of Fortey et at. 1995; see also Owens 1996), which is thus the age of P. bolbifrons. Diagnosis. Parabolinella with preglabellar field inflated into a balloon-like structure. Eye small, positioned anteriorly, and facial sutures moderately divergent both in front of and behind eyes. Description. Like most Parabolinella species this trilobite is of medium size. None of our material is complete, but incomplete dorsal shields are more than 70 mm long and whole animals certainly exceeded 80 mm in length. All specimens are somewhat flattened, and it is not possible to say much about the original convexity, other than that it is likely to have been low, as in the rare examples of the genus preserved in relief (see Henningsmoen 1957). The degree of flattening is likely to have affected certain features, for example, the angular divergence of the facial sutures. However, the preglabellar bulb is invariably preserved in some relief. None of the specimens extends posteriorly to the pygidium. Cephalic shield about two-thirds as long (sag.) as wide. Glabella rectangular, length hardly exceeding width, front notably truncate, occupying less than one-third of transverse cephalic width. The large cranidium on Plate 1, figure 1 shows a bulge in width at the level of IS. Occipital ring 20 per cent, of glabellar length, defined by deep occipital furrow which is effaced only at its extremities. Occipital tubercle present, but inconspicuous. Of glabella furrows, IS especially is marooned in glabella, the result of lateral effacement not uncommon in Parabolinella species; outer end forked. 2S appears to be longer, more complete on most specimens, curved inwards and backwards and extending to just over one-third of glabellar width. On other Parabolinella species 3S and 4S are very short and isolated within the glabella, and 3S at least can be seen on the specimen on Plate 1, figure 5. Palpebral lobes are short (exsag.), having about the same length as the occipital ring, are approximately in line with the outer end of S2 in a forward position, and are removed from glabella by about one-third of the width of its adjacent part. Posterior border furrow on fixed cheek shallow. Facial suture diverges in front of, and behind eye at a similar angle, but the length of the anterior section is one-quarter to one-third the length of the posterior section. Divergence varies between 30° to 55° to sagittal line, but this angle may well have been affected by flattening. Anterior border furrow is comparatively deep, and was probably originally deeper than lateral and posterior border furrows. The best preserved specimens show seven to ten FORTEY AND OWENS: BUBBLE-HEADED TRILOBITES 457 pits in the cranidial anterior border furrow to either side of the median inflation. This 'bulb' is a strongly inflated, subspherical structure which impinges on and grows over the border. Its transverse width is equal to, or slightly exceeds that of the glabella behind. Its posterior edge is usually marked with a few faint rugae, parallel to sagittal line. The dorsal surface lacks any sculpture. Free cheeks are yoked. Narrow lateral border is prolonged into long and very gently curved, needle-like, genal spines which seem to extend alongside the whole length of the body. A ventrally preserved pair of cheeks (PI. 1, fig. 4) shows that the doublure widened slightly, but distinctly across the median part having the bulb dorsally. On the largest specimen (PI. 1, fig. 3) doublure preserved on the right is seen to extend about half- way across the preglabellar field. Assuming it was bounded at its outer edge by the median section of the facial suture, this proves that the doublure underlay only part of the bulb. Although the hypostome is not preserved it is likely to have sat beneath the frontal lobe of the glabella (Text-fig. 2) as it is in the similar olenid Hypermecaspis (see Fortey 1990, pi. 1, fig. 1) and in other natant trilobites. In all other olemds known to us the doublure is extremely narrow and retains the same width across the mid-line. Thus the widening in P. bolbifrons is probably connected with a partial ventral extension of the bulb, but falls well short of the anterior hypostomal margin. No specimen is entire. There were 17 or 18 thoracic segments in total and presumably a comparatively small pygidium. The thorax is twice as long as the cephalic shield. Axis tapers very gently along the whole length of the thorax; half rings extend to almost half sagittal length of a ring. Pleurae are wider (tr.) than axis, and the pleural furrow extends almost along the length of each pleura as preserved, almost to spinose tips. Remarks. The bulb distinguishes the new species from all others of Parabolinella. However, there is some evidence of preglabellar inflation in some of the Argentine specimens attributed to Parabolinella argentinensis by Harrington and Leanza (1957, fig. 37.5). Fortey and Owens (in Owens et al. 1982, pi. 1, fig. g) figured a specimen attributed to Parabolinella argentinensis from the Tremadoc of South Wales which also seems to show a similar glabella inflation; although this specimen is seriously flattened it shows proportions comparable to those of P. bolbifrons. However, the bulb does not grow across the anterior border as it does in P. bolbifrons , and it is unlikely they are conspecific. text-fig. 3. Bulbolenus bellus Xiang and Zhang, 1983 ; reconstruction of cranidium for comparison with that of Parabolinella bolbifrons sp. nov. (based on Xiang and Zhang, 1983, pi. 33, figs 1, 3, 5); x8 approxi- mately. We note that there are similar bulbs developed on several Cambrian genera, for example, Amzasskiella Poletaeva, 1960 (= Triplacephalus Lu and Qian in Wang and Jin, 1977) and Nepea Whitehouse, 1939. However, the glabella structure, genal spines, and features of the thorax all indicate that P. bolbifrons is an olenid. We do not consider the possession of a bubble-head alone as an adequate basis for the erection of a new genus, for in almost all its other morphological features P. bolbifrons is very similar to Parabolinella triarthra (Callaway) (see Lake 1913, p. 68, pi. 7, figs 4-12; Fortey and Owens in Owens et al. 1982, pi. 1, fig. k). We regard the inflated lobe as no more than an autapomorphy of P. bolbifrons. A small exoskeleton in the Bristol collections associated with P. bolbifrons (BRSMG Cc2140) is 15 mm long and hardly shows any sign of median inflation of the preglabellar field. It is possible that the inflation only developed late in ontogeny (although the specimen is very likely to have been an holapsis); alternatively, this is a specimen of P. triarthra. Another olenid (or olenid-like) genus, Bulbolenus Xiang and Zhang, 1983 from the Cambrian of Xinjiang, China, has a swelling on the median part of the anterior border and anterior border 458 PALAEONTOLOGY, VOLUME 40 furrow, but this does not involve the preglabellar field, as in P. bolbifrons (cf. PI. 1, figs 1-8 with Xiang and Zhang 1983, pi. 33, figs 1-6 and pi. 34, figs 1-11, and Text-fig 1a and Text-fig. 3); therefore another, quite different type of anterior median swelling developed independently within the olenids. Finally, it may be considered possible that the bubble is a teratology, caused by infection by a parasite. For example, among the living fauna deformities in infected fish and arthropods are caused by parasitic copepods. We think that this is unlikely for two reasons. 1. The consistent size and position of the inflation. All our examples seem to be developed to the same degree, overhang the border to a comparable extent and show similar inflation. It is improbable that an infestation would always express itself in exactly the same fashion. 2. The form appears to be stratigraphically limited to a distinct population. If a parasite was the cause, one might with reason expect to find occasional pathological individuals in a range of different Parabolinella species, which is not the case. Acknowledgements. We thank Dr P. R. Crowther, formerly of the Bristol City Museum, and Dr E. Loeffler, Department of Geology, Bristol University, for loan of material described in this paper, and Mrs L. C. Norton for preparing the text-figures. Dr A. W. A. Rushton kindly drew our attention to Bulbolenus. REFERENCES angelin, N. p. 1854. Palaeontologica Scandinavica I : Crustacea formationis transitionis. Fasc. II. Lund, i-ix, 21-92, pis 25-41. brogger, w. c. 1882. Die silurischen Etagen 2 und 3 im Kristianiagebeit . &c. Universitats - programme (Kristiania), 376 pp., 12 pis. bruton, d. l. 1976. The trilobite genus Phillipsinella from the Ordovician of Scandinavia and Great Britain. Palaeontology , 19, 699-714. burmeister, H. 1843. Die Organisation der Trilobiten. Georg Reimer, Berlin. 148 pp., 4 pis. Curtis, M. l. k. 1968. The Tremadoc rocks of the Tortworth inlier, Gloucestershire. Proceedings of the Geologists' Association , 79, 349-362. fortey, r. a. 1990. Ontogeny, hypostome attachment and trilobite classification. Palaeontology , 33, 529-576. — harper, d. a. T., ingham, J. k., owen, a. w. and rushton, a. w. a. 1995. A revision of Ordovician series and stages from the historical type area. Geological Magazine , 132, 15-30. - and owens, r. m. 1990. Trilobites. 121-142. In McNamara, k. j. (ed. ). Evolutionary trends. Belhaven Press, London, xviii + 318 pp. - 199 In. The early Ordovician trilobite Beltella. Proceedings of the Bristol Naturalists' Society , 49, 69-79 [for 1989]. - 19916. A trilobite fauna from the highest Shineton Shales in Shropshire, and the correlation of the latest Tremadoc. Geological Magazine , 128, 437-464. — 1992. The Habberley Formation, youngest Tremadoc of the Welsh Borderland. Geological Magazine , 129, 553-566. Harrington, h. .1. and leanza, a. f. 1957. Ordovician trilobites of Argentina. University of Kansas Press, Lawrence, Kansas, Department of Geology Special Publication 1, 276 pp. hawle, i. and corda, a. j. c. 1847. Prodrom einer Monographic der bohmischen Trilobiten. Abhandlungen der Koniglichen Bohmischen Gesellschaft der Wissenschaften , 5, 1-176, pis 1-7. henningsmoen, G. 1957. The trilobite family Olemdae. . Shifter av det Norske Videnskaps-Akademie i Oslo , 7, Mat.-nat. Klasse for 1957 , 1, 1-303, pis 1-31. holloway, d. j. 1980. Middle Silurian trilobites from Arkansas and Oklahoma, U.S.A. Part 1. Palaeontographica , Abteilung A , 170, 1-85, pis 1-20. — and Campbell, k. s. w. 1974. The Silurian trilobite Oncyopvge Woodward. Palaeontology , 17, 409-422. hutchinson, r. d. 1962. Cambrian stratigraphy and trilobite faunas of southeastern Newfoundland. Bulletin of the Geological Survey of Canada, 88, i-ix-l- 1 160, pis 1 -25. ingham, j. K. 1974. The Upper Ordovician trilobites from the Cautley and Dent districts of Westmorland and Yorkshire. Part 2. Monograph of the Palaeontographica/ Society, 128 (538), 59-87, pis 10-18. jell, p. a. 1985. Tremadoc trilobites of the Digger Island Formation, Waratah Bay, Victoria. Memoirs of the Museum of Victoria, 46, 53-88, pis 19-33. FORTEY AND OWENS: BUBBLE-HEADED TRILOB1TES 459 lake, p. 1913, A monograph of the British Cambrian trilobites, part 4. Monograph of the Palaeontographical Society, 66 (324), 65-88, pis 7-10. lane, p. d. 1971. British Cheiruridae. Monograph of the Palaeontographical Society , 125 (530), 1-95, pis 1-16. lu yan-hao and chang weng-tang. 1974. [Ordovician trilobites.] 124-136, pis 49-56. In [Handbook of stratigraphy and palaeontology in southwest China.] Science Press, Beijing, iii + 454 pp., 202 pis. [In Chinese], MOORE, R. c. (ed.). 1959. Treatise on invertebrate palaeontology. Part O. Arthropoda 1. Geological Society of America and University of Kansas Press, Lawrence, Kansas, 560 pp. opik, a. a. 1967. The Mindyallan fauna of north-western Queensland. Bulletin of the Bureau of Mineral Resources, Geology and Geophysics, Commonwealth of Australia, 74, 1-404. owen, a. w. 1980. The trilobite Tretaspis from the upper Ordovician of the Oslo region, Norway. Palaeontology, 23, 715-747. owens, R. m. 1996. Trilobites of the Bristol district. Proceedings of the Bristol Naturalists' Society, 54, 67-84 [for 1994], — fortey, R. a., cope, j. c. w., rushton, a. w. a. and bassett, M. G. 1982. Tremadoc faunas from the Carmarthen District, South Wales. Geological Magazine, 119, 1-38, pis 1-7. poletaeva, o. k. 1960. [New genera and species of Cambrian trilobites of western Siberia.] Trudy Sibirskogo Nauchno-Issledovatel' skogo Instituta Geologii, Geofiziki i MineraVnogo Syr'ya , 8, 50-76, pis 1-3. [In Russian]. pribyl, a. and vanek, J. 1981. Studie zur Morphologie und Phylogenie der Familie Otarionidae R. & E. Richter, 1926 (Trilobita). Palaeontographica, Abteihmg A, 173, 160-208, pis 1-9. rasetti, f. 1966. New Lower Cambrian trilobite faunule from the Taconic sequence of New York. Smithsonian Miscellaneous Collections, 148, 9, 1-52, pis 1-12. resser, c. E. 1937. New species of Cambrian trilobites of the family Conocoryphidae. Journal of Paleontology, 11, 39 — 42, pi. 7. roemer, f. a. 1843. Die Versteinerungen des Harzgebirges. Hahn, Hannover, 40 pp., 12 pis. ruedemann, r. 1934. Paleozoic plankton of North America. Memoirs of the Geological Society of America, 2, 1-141, pis 1-26. snajdr, m. 1990. Bohemian trilobites. Geological Survey, Prague, 265 pp. thomas, a. t. 1981. British Wenlock trilobites. Part 2. Monograph of the Palaeontological Society, 134 (559), 57-99, pis 15-25. wallerius, i. d. 1895. Under sokningar dfver zonen med Agnostus laevigatus i Vestergotland jcimte en inledande dfversikt af Vestergotlands samtliga Paradoxides-/ager. Uppsala University Dissertations, Lund, Sweden, 72 pp. wang xiao-feng and jin yu-qin (eds). 1977. [ Palaeontological atlas of central and southern China (1. Lower Palaeozoic ).] Geological Press, Beijing, 411 pp., 116 pis. [In Chinese], whitehouse, f. w. 1939. The Cambrian faunas of north-eastern Australia. Memoirs of the Queensland Museum, 11, 179-282, pis 19-25. whittard, w. f. 1934. A revision of the trilobite genera Deiphon and Onycopyge. Annals and Magazine of Natural History, Series 10, 14, 505-533, pis 15-16. woodward, h. 1880. Description of a new species of trilobite, Onycopyge liversidgei , from the Silurian of New South Wales. Geological Magazine, 7, 97-99. xjang li-wen and zhang tai-rong. 1985. Trilobita. 64-136. In [Stratigraphy and trilobite faunas of the Cambrian in the western part of northern Tianshan, Xinjiang]. Geological Memoir of the People's Republic of China, Ministry of Geology and Mineral Resources. Series 2, 4, i-ix + 1-243 [In Chinese, English Summary]. zhou zhi-yi and dean, w. t. 1986. Ordovician trilobites from Chedao, Gansu Province, North-west China. Palaeontology, 29, 743-786. R. A. FORTEY Natural History Museum Cromwell Road London, SW7 5BD R. M. OWENS Department of Geology National Museum of Wales Cardiff, CF1 3NP Typescript received 23 February 1996 Revised typescript received 6 October 1996 A REVISION OF THE LARGE LAGOMERYCID ARTIODACTYLS OF EUROPE by BEATRIZ AZANZA and LEONARD GINSBURG Abstract. Large lagomerycids are only known in Europe from the Orleanian of the Loire Basin (France). At least three forms are recognized: Ligeromeryx gen. nov. praestans, Heterocemasl sp. and Lagomerycidae gen. et sp. indet. The genus Lagomeryx is restricted to small European species. If it is hypothesized that apophyseal appendages originated only once among cervoids, then a hypothetical brachyodont ruminant with divergent, supraorbital appendages bearing a small, distal fork which was cast from time to time, could be considered to be not only the common ancestor of lagomerycids and cervids, but also of merycodontines. Nevertheless, there is substantial evidence that appendages were acquired several times, and the possibility that lagomercyids are an entirely independent clade among cervoids is postulated. Heterocemas was the most primitive lagomerycid, having forked protoantlers with a prevalence of ramification by sprouting. The move evolved forms acquired multibranched construction (Ligeromeryx) and later, palmation at the protoantler basis ( Stephanocemas and Lagomeryx). Small size, accompanied by a subsequent reduction of the protoantler size, could have been acquired secondarily by Lagomeryx , probably when the lagomerycids filled forest-browsing niches. The first ruminants provided with antler-like appendages appeared in Eurasia during the early Miocene. In the early Orleanian (MN 3), both lagomerycids and cervids ( Procervulus , Acteocemas) were present in Europe. Lagomerycids were rare and represented by large to very small forms. Two large species have been recognized for a long time, Ligeromeryx gen. nov. praestans (Stehlin, 1937) and ‘ Stephanocemas ’ elegantulus (Roger, 1904). The appendages of the latter show a coronet-like structure and a coarse surface and consequently this species has recently been removed from Stephanocemas and referred to a dicrocerine deer (Azanza and Menendez 1990; Azanza 19936). Ligeromeryx praestans , formerly placed with smaller species in the genus Lagomeryx Roger, 1904, was founded on the basis of only three appendage specimens coming from Chitenay. In recent decades, numerous appendage remains have been collected from other localities in the Loire basin, that allow us to undertake new research into the nature, growth and evolution of lagomerycine antler-like appendages. Some dental remains have also been attributed to large lagomerycids. A complete systematic revision of all this material is presented in this study. The controversial phylogenetic position of the Lagomerycidae among cervoids is discussed as are relationships within the group. Institutional abbreviations used in this work are as follows. MB, Musee de Blois, Blois, Orleanais, France; MNEINP, Museum National d'Histoire Naturelle, Paris, France; MO, Musee d'Orleans, Orleanais, France; MS, Musee de Savigneen, Savignee-sur-Fathan, France; NHMB, Naturhistorischen Museum, Basel, Switzerland. GEOFOGICAL SETTING In the Loire Basin, large lagomerycid remains have been found in various continental sands of Orleanian age (Stehlin 1907; Mayet 1908; Denizot 1927; Ginsburg 1972). The sands originate from the French Massif Central and are in the form of patches overlying the Beauce limestone plain. Material from Chitenay, apart from the specimens of the type series, is housed in the collections of the Musee de Blois and the Naturhistorischen Museum, Basel. A revised list of the faunal assemblage can be found in Ginsburg (1990). This fauna, slightly older than that of the German (Palaeontology, Vol. 40, Part 2, 1997, pp. 461—485, 2 pls| © The Palaeontological Association 462 PALAEONTOLOGY, VOLUME 40 locality of Wintershof-West (MN 3), was placed at the lowermost part of the MN 3 (Ginsburg 1989, 1990; de Bruijn et al. 1992). In Anjou, similar sands were deposited on the pre-existing Esvres syncline and later reworked by the first transgression of the Falun Sea. Some appendages and dental remains have been collected recently from several sites where rich mammal faunas can be found in situ (Text-fig. 1): Les text-fig. 1. Geological sketch of the Loire Basin showing locations of the main sites from which large lagomerycid remains were examined. 1, continental sands, 2, marine Langhian Falun. Cht, Chitenay; D, Deneze-sous-le-Lude; P, Pontigne; Sa, Savigne-sur-Lathan. Beilleaux (parish of Savigne-sur-Lathan), La Brosse (parish of Deneze-sous-le-Lude) and Pontigne (quarry of Buissoneaux). This material is housed at the Museum National d’Histoire Naturelle, the Musee du Savigneen and some private collections. The associated fauna is listed in Ginsburg (1990). Apart from L. praestans , the species Lagopsis spiracensis, Steneofiber depereti janvieri, Xenoyus vendor and Andegameryx andegaviensis have never been found in the localities of the sands of Orleanais (MN 3b-A) nor in Wintershof-West. It seems that the sands of the Esvres syncline can be correlated with the sands of Chitenay. The Langhian fauna of Anjou contains the remains of contemporaneous mammals mixed with older ones, reworked from the underlying continental sands. According to Ginsburg (1990), these reworked sands are mainly of the same age as the fauna of Chitenay, but some specimens from Pont Boutard, at the eastern end of the same basin, suggest a younger age (MN 4), so we cannot exclude the possibility that some reworked specimens of the large lagomerycids described herein could be of a younger age. These specimens are housed at the Museum National d'Histoire Naturelle and Musee de Blois. AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 463 TERMINOLOGY The antler-like appendage of lagomerycids is composed of two components : the long, proximal one or ‘pedicle’ and the branched distal one. This distal part was capable of spontaneous autotomy in its entirety, as indicated by the rugosely, concave ventral surface observed in some specimens. Thus, casting of the distal part could occur from time to time, despite no coronet-like structure being formed. Therefore, it seems appropriate to name it ‘protoantler’, following A. B. Bubenik (1990). Protoantlers can branch by two mechanisms similar to those observed in deer antlers (A. B. Bubenik 1990). The first mechanism of branching is ‘splitting’, when the beam divides at the apex. However, they can also ramify through exostoses which form protuberances. Following A. B. Bubenik (1990), these cortical structures are termed ‘sprouts’ and the mechanism of ramification ‘sprouting’. In this study, the structures resulting from the ramification are termed ‘knobs’, ‘points’ or ‘branches’ according to their relative importance. SYSTEMATIC PALAEONTOLOGY Order artiodactyla Owen, 1848 Suborder ruminantia Scopoli, 1777 Infraorder pecora Linnaeus, 1758 Superfamily cervoidea Simpson, 1931 Family lagomerycidae Pilgrim, 1941 Genus ligeromeryx gen. nov. 1937 Lagomeryx Roger, 1904; Stehlin, p. 205, text-figs 10-12. Derivation of name. From ‘Liger’, the Latin name for the Loire river. All the material belonging to this ruminant comes from the Loire basin. Type species. Ligeromeryx praestans (Stehlin, 1937). Diagnosis. That of the species. Remarks. Individuals of this genus are larger than those of Lagomeryx. The pedicles bend forward and point outward more than in Lagomeryx or Stephanocemas. It also differs from Stephanocemas in its longer pedicles. The protoantler differs from that of Lagomeryx and Stephanocemas by the absence of a palmation, as occurs in Heterocemas , but the construction pattern is multibranched instead of forked. It differs from Lagomeryx also by a larger protoantler relative to the pedicle size and by the multibranched pattern, instead of being multipointed. Ligeromeryx praestans (Stehlin, 1937) Plates 1-2 1937 Lagomeryx praestans Stehlin, p. 205, text-figs. 10-12. Lectotype. NHMB/SO-3020, partial right frontal with the appendage preserved to the branch bases (Text-fig. 2d-e), figured by Stehlin 1937, fig. 10 as a syntype, and designated the ‘type specimen’ ( = lectotype) by Ginsburg et al. (1985). Paralectotypes. Two cast protoantlers (NHMB/SO-5720 and SO-2078; Stehlin 1937, figs I 1 -12). Diagnosis. A large lagomerycid in which the pedicles point outward in a plane very divergent to the sagittal one, and bend forward. The protoantler size is large relative to that of the pedicle. The 464 PALAEONTOLOGY. VOLUME 40 protoantler construction pattern is multibranched, without any true palmation being developed. The basic construction consists of three branches, two of which branch off closer together and generally more distally than the other. Commonly, there are accessory branches and knobs that modify this basic construction. Type locality. Chitenay, France (Lower Miocene, MN 3). Other localities. L. praestans has been found in situ in La Brosse and Les Beilleaux (Lower Miocene, MN 3) and reworked in several localities from Les Faluns (Middle Miocene, MN 5): Pontigne, Lasse, Deneze, Savigne, Noyant, Meigne-le-Vicomte, Chavaignes, Grand Trouve, Pont Boutard. All of these are placed in the Loire basin (France). Material. Apart from the type series, several mandible fragments are preserved from Chitenay. The material from La Brosse and Les Beilleaux comprises mainly dental remains; only a small protoantler fragment is preserved from Les Beilleaux. Among the appendage specimens from Les Faluns, are some frontal fragments which preserve the basal part of the pedicle, but the material mainly comprises protoantler specimens, none of which has branches preserved in their entirety. From Les Faluns there are also dental remains, generally isolated teeth. It must be pointed out that all dental remains found in the Loire basin are attributed here to L. praestans , usually the most abundant species, but the possibility cannot be excluded that some of them belong to another large lagomerycid. Measurements. See Tables 1-3. DESCRIPTION AND COMPARISONS Frontal bone and appendages The frontal bone is not well preserved. Only the upper region, from the orbital margin to the sagittal suture, remains. The supraorbital region is perforated by one supraorbital foramen without any depression being present. It is situated just anteromedially to the appendage base, being closer to the top of the frontal bone than in most primitive deer. The thick orbital rim in front of the appendages is nearly parallel to the sagittal plane. In cervids they converge anteriorly. The cranial appendages are supported completely by the supraorbital process of the frontal bone (without extending onto the braincase), as in all primitive deer. Thus, they are separated from the braincase, as in Lagomeryx , Procervulus ginsburgi and Acteocemas. Although the pedicles are vertically directed in lateral view, they point outward in a plane very divergent to the sagittal one (Text-fig. 2), more than in Lagomeryx and Stephanocemas. In most primitive cervids, they are parallel. The pedicle has a rounded cross section which can be flattened laterally in its distal part, just below the protoantler. It is noticeably bent anteriorly, whereas in Lagomeryx and Stephanocemas it slopes slightly posteriorly. It may also be slightly bend inwards in its distal part, as in Lagomeryx. Its surface is marked by very slight striations and by a deep groove that runs posteriorly to anteriorly on the medial side. It appears that there is very weak torsion. It is worth noting that the same morphology and disposition of pedicles also seem to be present in the Asiatic forms attributed to Lagomeryx , and in the problematical Heterocemas. The protoantler size relative to that of the pedicle is larger than in Lagomeryx. Its construction pattern is multibranched, rather than the multipomted pattern of Lagomeryx. The basic branches point approximately from the protoantler base but, in contrast with Stephanocemas and Lagomeryx , no true palmation is developed. The simplest protoantler morphology comprises three branches, as shown in the Chitenay specimens studied by Stehlin (1937). Their basal emplacements are situated approximately longitudinally according to the distal EXPLANATION OF PLATE 1 Figs 1-10. Ligeromeryx praestans (Stehlin, 1937). 1-2, MNHNP/Fs 3169; Pontigne; cast protoantler. Ldorso- external view. 2, internal view. 3-4, MNHNP/Fs 295; Pontigne; fragment of left protoantler. 3, dorsal view. 4, external view. 5-6, MNHNP/Fs 283; Pontigne; fragment of left appendage. 5, dorsal view. 6, external view. 7-8, MNHNP/Fs 1626; Clere-les-Pins; fragment of left protoantler. 7, dorsal view. 8, internal view. 9, MNHNP/Fs 2176; Deneze/La Brosse; fragment of left protoantler, external view. 10, MNHNP/Fs 285; Pontigne; fragment of cast protoantler, internal view. All x 1. PLATE 1 AZANZA and GINSBLIRG, Ligeromeryx 466 PALAEONTOLOGY, VOLUME 40 table I Dimensions (in mm) of the pedicle of large lagomerycids from the Loire Basin (France). L = maximal length; PAD = proximal anteroposterior depth; PTW = proximal transverse width; DAD = distal antero- posterior depth; DTW = distal transverse width. L PAD PTW DAD DTW Ligeromeryx praestans Chitenay NHMB/S.O. 3020 85-71 19-44 18-25 21-32 14-56 NHMB/S.O. 5720 — — — 17-65 11-91 NHMB/S.O. 2078 — — — 16-24 13-21 Pontigne MNHNP/Fs 298 — 12-77 12-6 — — MNHNP/Fs 294 — 17-88 18-24 — — MNHNP/Fs 296 — — — 19-62 10-92 MNHNP/Fs 301 — — — 19 1 16-11 MNHNP/Fs 300 — — — 15-21 12-49 MNHNP/Fs 285 — — — 21-65 10-22 MNHNP/M 3704 — — — 15-24 12-64 MNHNP/M 3162 — — — 14-82 11-35 MNHNP/MD 12 — 21-98 18-52 — — Clere-les-Pins MNHNP/Fs 1626 — — — 20-25 18-59 Auverse MNHNP/M 4567 — 19-85 21-27 — — MNHNP/M 4571 — — — 17-62 12-02 MNHNP/M 4802 — — — 16-56 11-11 MNHNP/M 4801 — — — 14-8 9-89 Grand Trouve MNHNP/MD 2 — — — 21-14 13-98 MNHNP/M 4133 — — — 23-97 16-19 Noyant-sous-le-Lude MNHNP/M 4135 — — — 12-62 8-2 MD 11 — — — 16-41 11-79 MNHNP/M 3339d — — — 16-82 12-41 MNHNP/M 4134 — — — 19-62 12-53 Deneze/La Brosse MNHNP/Fs 1609 — — — 15-93 12-68 MNHNP/Fs 2176 — — — 18-98 14-42 Lasse/Pont Brault MNHNP/Fs 1396 83 15-41 14-64 13-38 12-18 MNHNP/Fs 1395 — — — 18-93 14-34 Pont Boutard MNHNP/M 3222 — — — 20-34 12-12 Chavaignes MNHNP/Fs 5936 — — — 21-41 9 12 Heterocemas ? sp. Pontigne MNHNP/M 331 5g — — — 16-99 12-11 MNHNP/Fs 304 — — — 14-72 10-39 Meon MNHNP/Fs 6414 — — — 15-69 12-13 Auverse MNHNP/M 4572 — - — 17-27 11-66 MNHNP/M 4569 — — — 17-98 16-05 Pont boutard MNHNP/Fs 3914 — — — 18-13 14-99 Pontlevoy MNHNP/FP 3217 — — — 16-71 14-32 NHMB/CB1 179 — — — 23-18 15-58 467 AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS text-fig. 2. Protoantler terminology and orientation used in Ligeromeryx praestans (Stehlin, 1937). a. NHMB/SO-5720, paralectotype; Chitenay; cast protoantler, dorsal view, b, MNHNP/MD2; Grand Trouve; protoantler fragment, dorsal view, c, MNHNP/M-3222; Pont Boutard; protoantler fragment, dorsal view. d-e, NHMB/SO-3020, lectotype; Chitenay; right frontal with the appendage, d, dorsal view. E, anterior view. a = anterior, ace = accessory, pi = postero-internal and pe = postero-external branches, k = knob. Scale bar represents 20 mm. compression plane of the pedicle; this plane converges forward with the sagittal one, as can be seen in NHMB/SO-3020 (Stehlin 1937, hg. 10). The branches do not point equidistally. Two of them branch off closer together either more distally (morphotype A), or approximately from the protoantler base (morphotype B). In NHMB/SO-3020 (Text-fig. 2d) the basal branch is situated anteriorly (a) pointing inwards, and the two distal branches posteriorly, one pointing inwards (pi) and the other outwards (pe). Similar orientation can be recognized in MNHNP/Fs-283 (PI. 1, fig. 6) despite its four branches. Thus, it would seem logical to suppose that this orientation is the general condition. According to this interpretation, the orientation of NHMB/SO- 5720 is the opposite to that suggested by Stehlin (1937, fig. 11). It is noteworthy that Stehlin’s specimens show great variability in (1) the relative size and morphology of the branches; (2) the disposition of the branches, which range from nearly horizontal to vertical; and (3) the distance between the basal and distal forks. Moreover, the number and position of accessory branches, points or protuberances must be added to obtain a picture of the enormous morphological variability found in the protoantler construction of this species. Hence, the same horizontal branch disposition of Stehlin’s specimen NHMB/SO-5720 (morphotype B) is observed in the Grand Trouve (MNHNP/MD2) and Pont Boutard (MNHNP/M3222) specimens, despite the differences in relative size and morphology of their basic branches (Text-fig. 2). Branch a is the largest, being curved in MNHNP/M3222 as in NHMB/SO-5720, but straight in MNHNP/MD2. In contrast, the smallest one is branch pi in NHMB/SO-5720, but branch pe in MNHNP/MD2. In MNHNP/M3222, both pi and pe branches are about the same size. Nevertheless, the protoantler morphology of most of our specimens is referable to the vertical construction of NHMB/SO-2078, Stehlin’s specimen (morphotype A), although the protoantler base is enlarged by the presence of, at least, an accessory point which may just be a knob as in MNHNP/Fs-295 or developed as a branch as in MNHNP/Fs-301 and MNHNP/Fs-283. Ontogenetic growth. The nature of lagomerycid appendages has been discussed over a long time. Because of their relatively smooth surface and the absence of a coronet, they were interpreted as permanent skin-covered appendages (Stehlin 1939; Pilgrim 1941; Simpson 1945; Crusafont 1952; 468 PALAEONTOLOGY, VOLUME 40 Young 1964; Leinders 1983). However, as indicated by A. B. Bubenik (1983, 1990) and Vislobokova et al. (1989), their microstructure shows clearly that they are a direct outgrowth from the frontal bone, like deer antlers, and the presence of cast specimens has been demonstrated (Ginsburg 1985). A perennial apophyseal appendage might grow in diameter and length by periodical apposition of new bony lamellae, like the pedicle of deer appendages (A. B. Bubenik 1990). In such a case, we would expect to find that all specimens, apart from yearlings, would show the same morphology with variability reduced to individual variation in size and certain proportions. Ontogenetic variability like this is found in merycodontids (Frick 1937; Voorhies 1969). However, the variability found in lagomerycid appendages exceeds this substantially. The variation is comparable to that of deciduous deer antlers, which change in size and complexity with age so that a lineal ontogenetic sequence can be designated. The variability shown by our material is so great that any lineal sequence can be proposed. The growth mechanism of lagomerycid appendages seems to be more complex, and to explain this, their nature must be investigated by comparison with the most closely related extant appendage, the deer antler. Deer antlers are deciduous structures, the cycle and growth of which are dependent on the rise and fall of different androgen segregations, of which testosterone plays a dominant role (G. A. Bubenik 1990), therefore they develop in males in close relationship with their reproductive cycle. Nevertheless, if testosterone is substituted by some adrenal androgen, antlers can develop in both sexes. This seems to be the case of Rangifer (A. B. Bubenik 1975) and that of the Miocene Dicrocerus as we hypothesized (Ginsburg and Azanza 1991 ). It is possible that only the males of lagomerycids were provided with cranial appendages, as can be assumed from the complete skeletons found in Shanwang (China) It can be inferred that the role of testicular androgens in appendage development must be important, as is the rule in deer. After growth is complete, the deer antler mineralizes throughout, induced by a sudden rise of testosterone secretion. The blood supply to the surface is cut off and the tissues above the pedicle die; simultaneously a compact bridge between antler and pedicle is built up (A. B. Bubenik 1983, 1990). As soon as the testosterone levels approach the minimum the bridge is demineralized and a narrow zone of bone at the junction of the living bone of the pedicle and the dead bone of the antler is simultaneously destroyed by numerous osteoclasts (Goss 1970). The points of attachment between the antler and the pedicle are so attenuated that the weight of the antler itself effects the detachment. The base of a shed antler shows numerous spicules of bone that are remnants of the osteoclastic erosion (Goss 1970, 1983). The hypothesis that the distal part of the lagomerycid appendage could have been spontaneously rejected is supported by the fact that the ventral surface of some protoantler specimens is concave and shows these spicules (PI. 2, fig. 1). However, radiographs and longitudinal sections of these specimens (PI. 2, fig. 2) show that their rejection was produced without the protective bridge at the joint with the pedicle, as noted by A. B. Bubenik (1990). Indeed, the mineralization was not sufficient to cut oft' the blood supply from the pedicle and consequently the protoantler tissues were still alive when their rejection occurred. A similar casting process of tines or distal parts has been observed in the antlers of castrated deer (A. B. Bubenik et al. 1990). The lagomerycid protoantler was rejected in its entirety; only one specimen could be interpreted as a cast partial protoantler. EXPLANATION OF PLATE 2 Figs I -9. Ligeromeryx praestans (Stehlin, 1937). 1-2, MNHNP/Fs 302; Pontigne; cast protoantler. I, ventral view; x 1-3. 2, longitudinal section; x F5. 3-6, MNHNP/Fs 1294; Lasse; protoantler fragment. 3, lateral view showing a knob indicated by the arrow; x 1. 4, longitudinal section of the knob; x 7. 5, detail of the cortex-centre transition (Cr: centre, Cx: cortex); x 16. 6, detail of the knob; x 16. 7-9 transversal sections of specimen MNHNP/Fs 295 (in PI. 1, fig. 4). 7, section under the ramification; x 16. 8, section at the base of pe-pi branches; x 16. 9, detail of the centre part of the section of fig. 8; note the presence of secondary Haversian systems; x 52. PLATE AZANZA and GINSBURG, Ligeromeryx 470 PALAEONTOLOGY, VOLUME 40 MORPHOTYPE A MORPHOTYPE B text-fig. 3. Hypothetical ontogenetic growth of L. praestans protoantler combining two mechanisms: the protoantler casting and its subsequent regrowth by beam splitting (from top to bottom) and the cortical growth by sprouting (from middle to sides), a-f protoantler specimens corresponding to morphotype A. a, MNHNP/M4135; Noyant. B, MNHNP/M4134; Noyant. c NMB/SO2078, paralectotype; Chitenay. d, MNHNP/Fs 295; Pontigne. e, MNHNP/M4133; Grand Trouve. F, MNHNP/Fs 301; Pontigne. g-m protoantler specimens corresponding to morphotype B. G, MNHNP/Fs 3169; Pontigne. h, NMB/SO 5720, paralectotype; Chitenay. i, MNHNP/MD2; Grand Trouve. J, MNHNP/M-3222; Pont Boutard. k. MNHNP/Fs 1626; Clere-les-Pins. L, MNHNP/Fs 2176; Deneze. m, MNHNP/Fs 1395; Lasse or Pont Broult. Scale bar represents 30 mm. The mature deer antler is made up of an outermost region of compact bone containing a Haversian system, and a central region of spongy bone formed by fewer, coarser lamellae with wider marrow spaces. Secondary and tertiary Haversian systems and interstitial lamellae are absent in deer antler, presumably because the life of the antler bone is limited and the antler is laid down annually in its entire width from the beginning (Chapman 1975). The lagomerycid protoantler is constructed of rather immature compact bone. Although the core is more porous than the cortex, spongious bone trabeculae typical of the antler core are not developed (PI. 2, figs 8-9). The lamellae of the osteons of the cortex were not oriented in any particular direction in transverse sections of MNHNP/Fs-295 (PI. 2, figs 7-8), and do not confirm the presence of appositional lamellae supposed by A. B. Bubenik and figured in Cosoryx (A. B. Bubenik 1990, text-fig. 16a). However, in the longitudinal section of MNHNP/Fs-1394 (PI. 2, fig. 5), the lamellae are longitudinally oriented and a thin peripheral layer can be observed. Haversian osteons of secondary bone lamellae are observed mainly in the central region (PI. 2, fig. 9), but a dense Haversian tissue with several generations of Haversian systems, superimposed as in lifelong appendages (Rothschild and Neuville AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 471 text-fig. 4. Dentition of large lagomerycids from the Loire Basin, France, a-b, MNHNP/Fs 3580; Pontigne; right upper canine, a, lingual view, b, labial view. c-E, MNFINP/Fs 2397; La Brosse; left mandibular ramus, c, occlusal view. D, labial view, e, lingual view. All x 1. table 2. Dimensions (in mm) of the upper dentition of large lagomerycids from the Loire Basin (France). L = length; W = width. L W L W La Brosse Deneze MNHNP/Fs 2194 Ml/ 13-2 14-3 MNHNP/Fs 761 P2/ 12-9 10-2 MNHNP/Fs 2189 M3/ 142 15-8 MNHNP/Fs 2358 P4/ 9-4 1 1-2 Les Beilleaux Pontigne/Lasse BBX 63 P2/ 12-5 100 MNHNP/Fs 225 c/ 12-5 6-8 BBX 128 P4/ 1 14 10-0 MNHNP/Fs 228 c/ 12-8 61 BEI 534 Ml/ 150 14-8 MNHNP/M 200 C/ 12-7 6-5 BEI 315 M2/ 15-2 16-5 MNHNP/Fs 3980 C/ 12-7 6-6 BBX 285 M3/ 141 15-5 MNHNP/Fs 6159 C/ 12-7 6-3 Clere-les-Pms MNHNP/Fs 5361 P2/ 12-7 — MNHNP/Fs 1942 P4/ 9-7 121 MNHNP/Fs 3605 P4/ 110 12-4 Savigne MNHNP/Fs 3606 P4/ 9-3 12-5 MNHNP/Fs 2145 P3/ 11-7 10-3 MNHNP/Fs 5215 P4/ 9-4 1 2-2 Pont Boutard MNHNP/Fs 6143 P4/ 10-2 12-0 MNHNP/Fs 3867 M3/ 13-7 16-3 Noyant-sous-le-Lude MNHNP/Fs 908 P4/ 9-6 121 MNHNP/Fs 3023 M2/ 14 3 15-7 1910, pi. 6), is not developed. In contrast, it is developed in the pedicle bone. We conclude that protoantlers could be cast and regenerated without necessarily being annually deciduous. Deer antlers grow by proliferating fibroblast in their apices. These cells later become cartilaginous and are eventually incorporated into the bone trabeculae which strengthen the shaft (Goss 1970). So, beam-splitting (dichotomous branching at the tip) is the usual mechanism of ramification. However, they can ramify also through exostoses, which form sprouts. Sprouting is present in 472 PALAEONTOLOGY, VOLUME 40 text-fig. 5. Upper dentition of large lagomerycids from the Loire Basin, France, a, MNHNP/Fs 2145, Savigne; right P;!. b, BEI 534. Les Beilleaux; right M1. c, BBX 285, Les Beilleaux; right M:1, occlusal view. Scale bar represents 10 mm. Rangifer , in the first antler of Cervus elaphus , in the second antler of North American Alces and the prong in Odocoileus (A. B. Bubenik 1990). As described above, the protoantler of Ligeromeryx shows numerous protuberances of knobs that are cortical structures (PI. 2, figs 3-6). Possibly these knobs eventually developed as accessory branches. Some of them might have had a genetic basis, as they have frequently been found in the same position, but many others have not. It appears that sprouting might have been a very important process of ramification in lagomerycids, to judge by the frequency and versatility of the accessory branches and knobs. If these interpretations of the nature of the lagomerycid appendage are correct, protoantler growth in Ligeromeryx was influenced by (1) total or partial protoantler casting and its subsequent regrowth by beam splitting; and (2) cortical growth by sprouting or appositional lamellae. We hypothesize that, if the first mechanism occurred, the protoantler morphology would reproduce the basic pattern with three or perhaps four branches, but if casting did not occur, then the second mechanism would modify this basic pattern resulting in the enormous variability of Ligeromeryx morphology. Specimens that correspond to both the vertical and horizontal patterns are ordered in Text-figure 3 according to a hypothetical ontogenetic sequence that combines these two mechanisms. The youngest state is attributed to small three-pointed specimens, the morphology of which resembles that of the more adult specimens. Among the material from the Pontigne-Savigne Basin there are some unbranched specimens. MNHNP/M4800 from Les Beilleaux is a complete appendage that, in contrast with other slender pedicle fragments (MNHNP/Fs 298), is less divergent, right in frontal and lateral views and without any trace of torsion. In our opinion, it is not attributable to Ligeromeryx praestcms but to Lagomervx ruetimeyeri. A few peculiar specimens cannot be placed in this scheme. Their taxonomic position is discussed later. Upper dentition The upper canine (Text-fig. 4a-b) is very long, slender and curved downward and backward. Its anterior edge is thicker than the posterior one, where both the labial and lingual faces join bevelled to form a sharp ridge. The labial face bulges anteriorly while the lingual one is flatter and shows a weak longitudinal groove that runs along its posterior part from the base to within 18-31 mm of the apex, depending on the individual. This confers a sigmoidal profile on the lingual face. P2 is a long thin tooth. On the labial wall, the parastyle and the paracone protrude, but less than in Procervulus or Dicrocerus, and are joined basally. The labial side of the metacone is flatter and is separated from the paracone by a groove. On the lingual lobe, the slightly protruding protocone is in a central position as in Procervulus, but the tooth is much longer. P3 resembles P2 but both the parastyle and the paracone are less divergent on the labial wall. The protocone is more protruding on the lingual lobe than in P2 and the hypocone seems consequently to be slightly thrown back (Text-fig. 5a). According to this, the difference between P2 and P3 appears to be clear. However, study of the rich material of Dicrocerus elegans from Sansan leads to the conclusion that these features are variable and are sometimes reversed. Only teeth observed in situ on the maxillary can be identified with certainty. P4 is a short wide tooth. It is almost symmetrical, the protocone being approximately in a central position on the lingual lobe. The parastyle, paracone and metastyle protrude from the labial wall. A medial fold is present. The lingual cingulum is weak, sometimes reduced to a basal bulge. AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 473 table 3. Dimensions (in mm) and statistics of the lower dentition of large lagomerycids from the Loire Basin (France). L = length; W = width. P/2 P/3 P/4 M/1 M/2 M/3 L W L W L W L W L W L W Chitenay NHMB/SO 2060 — — 10-7 5-8 12-4 7-3 12-25 9-0 13 6 10-0 20-3 9-8 NHMB/SO 3027 7-8 3-7 9-7 5-3 10-65 6-4 — 8-5 13-05 9-6 19-2 9-5 NHMB/SO — — — — 11-4 6-8 — — — — — — MB/M 3199 8-85 3-9 — — — — — — — — — — MB/M 3193a 9-4 4 0 10-4 5-5 11-8 7-8 12-3 91 — — — — MB/M 3193b 12-9 9-75 19-9 9-85 MB/M 3193c — — — — — — — — 1 3-4 9-9 20-2 9-7 MNHNP/CHT4 13-6 8-9 — — n 3 3 3 3 4 4 2 3 5 5 4 4 min. 7-80 3-70 9-70 5-30 10-65 6-40 12-25 8-50 12-90 8-90 19-20 9-50 max. 9-40 4 00 10-70 5-80 12-40 7-80 12-30 9-10 13-60 10-00 20-30 9-85 mean 8-68 3-87 10-27 5-53 11-56 7-07 12-27 8-87 13-31 9-63 19-90 9-71 La Brosse n — — — — 2 2 3 3 2 3 2 3 mm. — — — — 11-40 6-75 11-70 8-70 12-70 8-70 17-90 9-50 max. — — — — 1 1 40 7-0 12-45 9-40 13-80 10-80 19-80 9-90 mean — — — — 11-40 6-88 12-20 8-97 13-25 9-62 18-85 9-70 Pontigne n — — 3 3 7 7 5 5 13 13 1 1 11 min. — — 10-10 5-20 11-40 6-85 11-90 8-00 12-00 7-40 18-60 8-80 max. — — 12-00 6-20 13-10 7-70 12-55 9-40 14-90 10-20 21-40 10-30 mean — — 10-98 5-58 12-80 7-21 12-31 8-77 13-61 9-57 19-51 9 61 Lasse n — — 1 1 — — — — — — 2 2 min. 18-80 9-40 max. 20-20 10-45 mean — — 11-10 5-40 — — — — — — 19-50 9-92 Lasse/Pontigne n — — 1 1 6 6 1 1 9 9 1 1 min. — — — — 11-50 6-60 — — 12-10 8-60 — max. — — — — 12-90 7-20 — — 14-60 10-40 — — mean — — 10-70 6 10 12-18 6-85 1 1-5 9-10 13-21 9-46 19-25 9-85 Pont Boutard n — — — — 1 1 — — 3 3 — — min. 13-10 9-25 — — max. 13-55 9-70 — — mean — — — — 11-70 7-00 — — 13-33 9-42 — — Deneze n — — — — 4 3 1 1 3 3 2 2 min. — — — — 11-60 7-30 — — 12-70 8-80 18-00 9-40 max. — — — — 12-70 7-60 — — 13-90 10-30 1915 9-95 mean — — — — 12-15 7-47 1 1-9 8-7 13-83 9-57 18-58 9-67 Noyant n — — 1 1 1 2 2 1 1 5 5 min. — — — — — — 1 1-40 9-70 — — 18-25 9-15 max. — — — — — — 12-10 10-00 — — 19-70 1000 mean 6-50 12-60 7-30 1 1-75 9-85 1 3-50 1000 19-01 9-76 474 PALAEONTOLOGY, VOLUME 40 text-fig. 6. Lower dentition attributed to Ligeromeryx praestans from the type locality, a-b, MB/M 3199; fragment of right mandible ramus with P2 and fragment of P3. a, occlusal view. B , lingual view, c-d, MB/M 3193a; left P.-Mj. c, occlusal view. D, labial view. Scale bar represents 10 mm. The upper molars have an approximately square outline. The external wall relief is strong, as in Procervulus and Dicrocerus , with the metastyle much more protruding than the metacone, but very weak in comparison with the parastyle, the paracone and the mesostyle. The anterior and posterior lobes are parallel but somewhat oblique to the longitudinal axis in M1 (Text-fig. 5b). The posterior lobe is moved slightly outwards in comparison with the anterior one. This character decreases from M1 to M3, reaching the same level. The lingual cones are developed, the protocone being more pronounced than the metaconule from M1 to M3. A very weak central fold-like structure is present only in M1. The postprotocrista ( = protoconal fold in Heintz 1970) is short in M1 but is more developed in M2 and M3 turning labially (Text-fig. 5c). The endostyle is in general strong and the development of the cingulum is variable. Lower dentition There is no P4 as in most cervoids (Text-fig. 6a). P2 has two roots (Text-fig. 6b. d) and is small, low, long and very thin. The paraconid is pointed and turned lingually. The crest coming down forward from the protoconid takes up a position approximately on the longitudinal axis. The anterior valley is shallow and broad. The metaconid, when present, is very small and attached to the postero-lingual side of the protoconid (Text-fig. 6c), A broad external groove is insinuated. P;! is long and thinner than in Procervulus and much more so than in Dicrocerus. The biggest specimens have the anterior elements differentiated ; the paraconid is well developed but close to the parastyle, so it is not visible if the tooth is moderately worn. The anterior valley is deep and wide. The metaconid is not individualized from the short oblique cristid (Text-fig. 6c). The entoconid and the entostylid are well developed and reach the postero-lingual corner. The external groove is weak, but deeper than in Dicrocerus. On the external wall, the hypoconid shows a basal bulge on the P3 of the mandible from Les Beilleaux. Pj is thicker and somewhat longer than in P3. However, it is shorter than in Dicrocerus and Procervulus because of the reduction of its anterior part. The anterior valley and the anterior crest of the protoconid are consequently shorter than in Dicrocerus and Procervulus. It is not molarized. The metaconid is individualized from the short oblique cristid and is usually almost opposite the protoconid and develops a short postero- lingual crest (Text-figs 4c, 6c). The metaconid on the P4 from Les Beilleaux and in some specimens from Les Faluns, protrudes on the lingual profile and is placed thrown back developing no postero-lingual crest. The entoconid is well developed, closing the posterior valley. The external groove is deep whilst the protoconid is very delimited on the external wall. The hypoconid may bulge toward its base in some specimens. In the lower molars, lobe disposition is variable, although in most cases they tend to be disposed obliquely. The lobes are bulging on the internal wall, whose relief is well developed, while the mesostylid is prominent (Text-fig. 4e). The relief is not very well developed on the molars of the mandible from Les Beilleaux and some specimens from Les Faluns, the mesostylid being less protruding than in Dicrocerus. The Palaeomeryx- fold is strong. The ectostylid and the cingulum are weaker from M4 to M3. The internal cristids are relatively long but never overlapping. A diagonal connection forms the interlobular union. The postmetacristid and the AZANZA AND G1NSBURG: LAGOMERYCID ARTIODACTYLS 475 prehypocristid tend to be joined to this connection. The third lobe of M,, is long and placed on the longitudinal axis so that there is an inflection on the lingual wall. Genus heterocemas Young, 1937 Heterocemasl sp. Material. Twelve appendage fragments from Meon, Pontigne, Pont Brault, Pont Poutard, Auverse, Chalonnes and Meigne-le-Vicomte, housed at the MNHNP. From Pontlevoy, MNHNP/Fp3217 and probably NFIMB/Bourgeois collection- 1 179 also belong to this form. Description and comparisons. The protoantler specimens included in this form correspond to a basic forked construction in which one branch (probably the posterior) is distinctly longer than the other, sometimes also having the tip forked. This morphology is clearly shown by MNFINP/Fs 6414 which is also sharply bent inward (Text-fig. 7a-b). The presence of knobs is the rule as in L. praestans. There is usually a knob on the text-fig. 7. Heterocemasl sp. from the Loire Basin, France, a-b, MNHNP/6414; Meon; left protoantler. A, anterior view, b, external view, c, MNHNP/Fs 304; Pontigne; left protoantler presumably juvenile, external view, d, MNHNP/M 331 5g ; Pontigne; left protoantler, external view. All x 1. 476 PALAEONTOLOGY, VOLUME 40 posterior branch situated on its lower part (Text-fig. 7b-d) or in the middle (in NHMB/1179, MNHNP/Fs 1391). Only MNHNP/M4569 from Auverse has sufficient preserved structure below the main fork to be certain that it is the pedicle. The size of these specimens is comparable to that of L. praestans and to Procervulus dichotomus, but the section of the protoantler just below the main fork is rounded instead of elliptical as in Procervulus (Text-fig. 8). MNHNP/Fs 304 (Text-fig. 7c) from Pontigne is very small and is presumably a juvenile protoantler. The text-fig. 8. Scatter plot of transverse/antero-posterior diameters of the protoantler (measured just below the main fork) of Heterocemas ? sp. versus Procervulus dichotomies from the Loire Basin. (*Pontlevoy specimen referred to Procervulus aurelianensis by Mayet 1908). Pontlevoy specimen NHMB/1179 (figured as Procervulus aurelianensis by Gaudry 1878, text-fig. 100c and Mayet 1908, text-fig. 94c) is much bigger and the anterior branch is also forked, and must probably belong to an old individual. These specimens cannot be considered as belonging to L. praestans, despite the versatility assumed for their protoantler morphology. Certainly there is a prevalence of branching by sprouting and the morphology seems to represent the extreme of morphotype A. Nevertheless, the forked and multibranched patterns can be clearly separated. In contrast, they resemble greatly the problematical Heterocemas simpsoni Young, 1937 and to a AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS All lesser extent Heterocemas gracilis (Vislobokova, 1983). In our view, it seems feasible that a form closely related to these Asian species was present in the Loire basin. Lagomerycidae gen. et sp. indet. Material. One distal fragment of protoantler (MB specimen) from Chitenay. Two fragments from Fay-aux- Loges belonging to the same cast specimen (MO/827), one of which was figured by Mayet ( 1908, pi. 4, fig. 17). Description and comparisons. The Chitenay specimen (Text-fig. 9) shows a very unusual multibranched pattern text-fig. 9. Lagomerycidae gen. and sp. indet. a-b, MB; Chitenay; distal fragment of the protoantler. A, dorsal view, b, lateral view. Scale bar represents 20 mm. not referable to any of the described lagomerycid forms. The protoantler is flattened, showing a tendency to form a vertical palmation, the distal border of which bristles with two ranges of branches or knobs. It is possible that this specimen is an aberrant protoantler, as is from time to time found in deer appendages. Nevertheless, it should be pointed out that this morphology resembles that of the merycodontid Ramoceros ( Merriamoceros) (Frick 1937, text-figs 35a, 40a), suggesting that it might really correspond to an as yet incompletely known form. Two other fragments of cast specimens show a flattened scar suggesting that they might also belong to this form. DISCUSSION As described above, we recognize at least three forms among the remains of large lagomerycids found in the Loire basin. The most abundant, Ligeromeryx praestans, was included for many years in Lagomeryx Roger, 1904. This genus was defined as small ruminants with antler-like appendages that are constituted by a long pedicle supporting a rather small protoantler built by a palmation surrounded by a crown of small points (multipointed construction). Its taxonomic status has been revised recently by Gentry and Heizmann (1993) who exposed the problem concerning the species type and asked the International Commission on Zoological Nomenclature to designate L. ruetimeyeri Thenius, 1948 as the type species, the holotype being the Reisenburg left appendage illustrated by Rutimeyer (1880, pi. 1, figs 2-3). Other smaller species included in the genus are L. parvuhts Roger, 1 904 and L. pumilo Roger, 1904. L. simplicicornis Schlosser (1904) was described as a Lagomeryx with unbranched appendages, nevertheless, the specimen illustrated by Schlosser (1904, pi. 26, fig. la) corresponds to a pedicle whose protoantler was cast (Antunes et al. 1994). Some other large Asian forms have also been referred to this genus but, as discussed below, the protoantler construction is not the same and it should be excluded from the genus. 478 PALAEONTOLOGY, VOLUME 40 Along with Lagomeryx , most authors have included Procervulus Gaudry, 1878 and Climacoceras Maclnnes, 1936 in the family Lagomerycidae Pilgrim, 1941. In contrast to lagomerycids, splitting of the beam is the predominant process of branching in the protoantler construction of the European Procervulus ; moreover, the protoantler is ornamented and the upper molars show a clear central fold. These features place Procervulus closer to true deer (Ginsburg, 1985; Azanza 19936). In Climacoceras , it appears that sprouting is the predominant process of branching of its appendages, although they have no differentiated pedicle and are perennial. Other dental and postcranial features placed Climacoceras closer to giraffoids (Hamilton 1978; Janis and Scott 1987). By contrast, a form very close to Lagomeryx is the Asian Stephanocemas Colbert, 1936, as noted by Ginsburg (1985). It comprises medium to large forms in which the morphology of the appendages resembles that of Lagomeryx. The protoantler is built also by palmation but is surrounded by a crown of branches (multibranched construction) instead of points. Moreover, the pedicle is relatively short (at least in the type species) whilst the protoantler is very large. Along with the type species, Stephanocemas thomsoni Colbert, 1936 from Tung Gur, we recognize S. tsaidamensis Bohlin, 1937 (including the material described by Bohlin as Cervidae sp.), S. aralensis Beliajeva, 1974, and 5. rucha Ginsburg and Ukkakimapan, 1983. Two European species have been included in this genus for a long time, Acteocemas infans (Stehlin, 1939) and Stehlinoceros elegantulus (Roger, 1904). They have a coronet-like surrounding to the protoantler base and the surface is ornamented, so they are considered to be closer to Dicrocerus (Azanza 19936). Paradicrocerus flerovi Gabounia, 1959 was described on the basis of only one specimen (Gabounia 1973, pi. 8, fig. 3) showing a multibranched construction resembling that of Stehlinoceros elegantulus. This could be an aberrant specimen of Dicrocerus , also represented in Belometchescaya, exhibiting a construction similar to that found in extant Muntiacus. Nevertheless this morphotype is not present in the rich population of Dicrocerus from Sansan. Moreover, a cranium that Gabounia (1973, pi. 8, fig. 2) illustrated as Dicrocerus sp. belongs to P. flerovi. It shows short divergent pedicles very distant from each other and the supraorbital foramen is very close to the frontal roof. This morphology is not present in Dicrocerus but is in Stehlinoceros elegantulus. It is possible that Stehlinoceros Azanza and Menendez, 1990 is a junior synonym of Paradicrocerus. The large French lagomerycid Ligeromeryx praestans differs from both Lagomeryx and Stephanocemas because there is not true palmation at the protoantler basis. Its protoantlers are multibranched, as in Stephanocemas , but not multipointed as in Lagomeryx. Moreover, the size proportion between the protoantler and the pedicle is bigger than in Lagomeryx. These features are also shared by some Asian species referred either to Stephanocemas or to Lagomeryx. They are L. triacuminatus (Colbert, 1936) and L. colberti (Young, 1937) ( = L. teilhardi Young, 1964). They show a more complex morphology of protoantlers but their dentition is more primitive, with still preserved (Chow and Shih 1978; Vislobokova et al. 1989). It seems that it could be related to Ligeromeryx but this matter needs further study. It is worth mentioning that the protoantlers of the problematical Heterocemas Young, 1937, resemble those of Ligeromeryx in the absence of a palmation but their construction is not multibranched but rather forked. Vislovokova (1983) included Heterocemas in Procervulus , but the former has very divergent curved pedicles and the surface of the appendage is smooth, as in lagomerycids. As described above, some incomplete specimens found also in the Pontigne-Savigne Basin show a similar morphology but the first branch is more reduced and the presence of knobs seems more predominant. Probably, the enigmatic specimen from Pontlevoy NHMB/1 179 (figured as Procervulus aurelianensis by Gaudry 1878 and Mayet 1908) belongs to this form. Although larger and with a more complex morphology, this specimen could be placed into a sequence of ontogenetic development comparable to that hypothesized for L. praestans. These specimens might belong to Heterocemas but the material is not sufficiently well preserved to be certain. Finally, we note the possibility that there is a greater diversity of lagomerycid forms among our material. Despite the versatile construction that we assume for the protoantler of Ligeromeryx , the peculiar specimens described as Lagomerycidae gen. et sp. indet. cannot be feasibly ascribed to it. The material is too incomplete to lead to any conclusions. AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 479 There is also great variability in the dentition attributed to large lagomerycids. The mandible from Les Beilleaux described by Ginsburg et al. (1985) is bigger than that of Chitenay and La Brosse; the premolars are longer and thinner (Text-fig. 10) and with a relatively simple morphology. LP/2 LP/3 LP/4 LM/1 LM/2 LM/3 WP/2 WP/3 WP/4 WM/1 WM/2 WM/3 text-fig. 10. Comparative measurements of the lower dentitions of large lagomerycid from the Loire Basin. (100 = Procervulus ginsburgi from Artesilla, Spain). The molars are also long with the inner cristids in line and a weak metastylid. By contrast, the morphology of the teeth from Chitenay and La Brosse is more like that of cervids. A slight difference of age has been argued to explain these differences (Ginsburg 1990). PHYLOGENETIC RELATIONSHIPS The lagomerycids have been one of the most controversial ruminant groups because of the different interpretations about the nature of their appendages. They have been considered to be aberrant giraffoids, either a separate lineage or a junior synonym of the Palaeomerycidae (Pilgrim 1941; Simpson 1945; Young 1964), or cervoids. Their cervoid affinities now seem to be firmly established by cranial and postcranial features (Chow and Shih 1978; Leinders and Heintz 1980; Vislobokova et al. 1989) but there is no consensus over the phylogenetic position within that group. Thus, they have been considered to be a group (1) that represents the perennial stage or the ‘pre-antler stage' in the evolution of antlered cervids (Crusafont 1952; Leinders 1983; Gentry 1994); (2) included into the family Cervidae either as a separate subfamily (Vislobokova et at. 1989) or as a junior synonym of Muntiacinae (Chow and Shih 1978); (3) more closely related to antilocaprids (Ginsburg 1985; Solounias 1988); or (4) that represents a possible independent clade (G. A. Bubenik and A. B. Bubenik 1986; Azanza 19936). The fact that Procervulus , the most primitive cervid (Ginsburg 1985; Azanza 19936), has been included among lagomerycids for a long time, and even its synonymy with Heterocemas proposed (Vislobokova 1983), demonstrates the great resemblances between the procervuline and lagomerycid 480 PALAEONTOLOGY, VOLUME 40 protoantlers. Both appendages show: (1) long upright pedicles above the orbits; (2) absence of a coronet (or any structure resembling one) and no evidence of ‘velvet’ shedding, in contrast with other primitive deer lineages such as the dicrocerines (Azanza 1993/?); (3) presence of cast specimens indicating occasional protoantler rejection, which occurred presumably when the tissue was still alive. We noticed that the procervuline protoantler has a remarkably ornamented surface and its predominant process of branching is by splitting of the beam tip (Ginsburg 1985; Azanza 19936). These features indicate that, as in true antlers, growth occurs at the tip and presumably requires a more intense vascularization of the ‘velvet’ than in the lagomerycid protoantler (Azanza 19936). In contrast, the most important feature of the lagomerycid protoantler is the predominance of branching by sprouting which translates into enormous versatility of protoantler construction. Branching by sprouting indicates a highly active cortex and could be linked to the protoantler structure because mineralization progresses centrifugally (A. B. Bubenik 1990). It must be pointed out that in dicrocerines the mineralization is clearly centrifugal, but nevertheless the presence of sprouts is a rarity. Sprouts are hardly ever present in primitive procervulines (Azanza 1993«), as well as in some extant deer (A. B. Bubenik 1990). Azanza (19936) emphasized the prevalence of branching by sprouting in the growth of lagomerycid appendages and considered this feature to be a useful synapomorphy to define this family. Although belonging to cervoid ruminants, Ginsburg (1985) considered the Lagomerycidae more closely related to the Antilocapridae. This argument is based on the great similarity in appendage construction between one of the most ancient antilocaprids, the merycodontine Ramoceros, and Ligeromeryx praestans. Both taxa have a similar three-branched structure of the protoantler and the pedicles are divergent, long and inwardly curved. Concerning dental features, the upper molars lack the central fold and the lingual cingulum is absent or very weak. Despite these resemblances, merycodontines differ from lagomerycids in the following features. 1. A simple ontogenetic sequence: small yearling appendages and adult specimens have the same morphology, the variability being reduced to an individual variation in size and certain proportions (Frick 1937; Voorhies 1969). 2. The total absence of sprouts. 3. The common presence of one or several pseudocoronets that are not homologous to the coronets of antlers. They can be asymmetrical in both appendages of the same individual and can be developed either over the pedicle or over the branches. This structure has been variously interpreted. Voorhies (1969) suggested that it is related to a periodic regression of the skin anticipating the casting of the horn sheath in Antilocaprinae. He argued that the skin would have been present only during the period of additive growth and regenerated over the whole, bare and dead appendage but this is difficult to accept (A. B. Bubenik 1990). According to A. B. Bubenik (1990), they might have evolved when the distal part of the appendage was sequestered, or at a stage to be lost as a situation similar to that known in deer prior to the velvet shedding. This interpretation is surprising when it is taken into account that this structure is the rule and the cast appendages the exception. 4. Little evidence of casting. According to A. B. Bubenik (1990), a few pedicles with a bare surface above the uppermost pseudocoronet exit (e.g. the right appendage of F: A.M. 32895 figured by Frick 1937, fig. 27). Surprisingly there is no evidence of cast protoantlers. 5. Unenlarged upper canines occasionally retained, hypsodont cheek teeth, Paleomeryx- fold and metastylid lost, complete postentocristid. 6. Lacrimal depression absent, nasals and muzzle extremely enlarged, inflated auditory bullae, lateral metacarpal partially retained occasionally (Frick 1937). It could be argued that these differences are autapomorphies and do not exclude a closer relationship between them. The origination of merycodontines may have occurred by geographical speciation. This conspicuous speciation event could have taken place when the ancestors migrated from Eurasia during the latest early Miocene. Prior to this date no evidence of merycodontines or ancestral taxa has been found in the North American palaeontological record. Merycodontines AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 481 quickly acquired hypsodonty, as well as the other characteristic cranial and postcranial features; their descendants acquired the horn sheath. It must be pointed out that during the mid Miocene hypsodonty was acquired by other groups like the equids in North America while their Eurasiatic counterparts remained brachyodont. A hypothetical brachyodont ruminant provided with divergent, supraorbital appendages whose rather small, distal fork could or could not be cast from time to time, is considered tentatively as a common ancestor of both groups and also to cervids. This was inferred from the resemblances of the appendage construction of the most primitive representatives of each group: Paracosoryx, Heterocemas and Procervulus. Apart from the forked, occasionally deciduous protoantler, no apomorphy is shared among them (Text-fig. 11). In addition, the differences should not be text-fig. 11. Morphostructural features and physiological processes of the different antler-like appendages. overlooked on the above mentioned morphostructural features. These are correlated with differences in physiological processes suggesting that they correspond to separate types of protoantlers. Lagomerycids and procervulines, as well as dicrocerines ( Acteocemas ), appeared in Europe during the early Miocene, MN 3 (Text-fig. 12). The first record of merycodontines is text-fig. 12. Biochronological distribution of the antler-like appendages. 482 PALAEONTOLOGY, VOLUME 40 Paracosoryx (‘ Merycodus ' prodomus Cook, 1934) in the uppermost Arikareean (Tedford et al. 1987) correlated with the lower part of the MN 3 (Steininger et al. 1985). At precisely the same time, frontal appendages were developed in different lineages of artiodactyls, induced by the onset of marked seasonality (Morales et al. 1994). In this context, the independent evolution of protoantlers in each group seems feasible. As discussed above, we recognize four protoantler morphologies typifying four genera: Heterocemas , Ligeromeryx, Stephanocemas and Lagomeryx. The hypothesis of phylogenetic relationships among them is illustrated by the cladogram of Text-figure 13. In our view, it is feasible LAGOMERYCIDAE W H ^ that the most primitive morphology of the lagomerycid protoantler was a forked construction with a prevalence of ramification by sprouting. The protoantler of Heterocemas seems to correspond well with this construction. All the other lagomerycids shared the presence of three or more branches, i.e. they acquired the multibranched construction as is preserved in Ligeromeryx. The more evolved forms ( Stephanocemas and Lagomeryx) acquired palmation at the protoantler basis. It seems feasible that the small size and the reduction of the protoantler size were acquired secondarily by Lagomeryx , so the multipointed construction could be related to it and be considered to derive from the multibranched construction. The reduction of size, accompanied by a subsequent reduction of the protoantler size, is a trend also found in the South American deer genera Mazama and Pudu , which have radiated to fill forest-browsing niches in a manner comparable to that shown by the forest duikers in Africa and the muntjaks of Asia (Eisenberg 1987). This interpretation could explain the almost total absence of small lagomerycids in the faunas from the Spanish central basins (Antunes et al. 1994), where a greater predominance has been detected of inhabitants of open AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 483 habitats than the contemporaneous faunas of the Valles-Penedes and other European basins (Alberdi et al. 1985). Acknowledgements. This work has been partly funded by the ‘Direction General de Investigation Cientifica y Tecnica’, project PB92-0066 ‘Historia evolutiva de los Rumiantes del Mioceno inferior y medio de Espana: contribution al conocimiento del origen y diversification de los Rumiantes con apendices craneales’. The authors are grateful to Dr Jorge Morales for his comments on a draft of the paper. We thank also Dr Marina Alcobendas for her useful observations on the protoantler bone. REFERENCES alberdi, M. T., hoyos, m., mazo, a., morales, j., sese, c. and soria, d. 1985. Bioestratigrafia, paleoecologia y biogeografia del Terciario de la provincia de Madrid. 99-105. In alberdi, m. t. (ed.). Geologia y paleontologia del Terciario continental de la Provincia de Madrid. C.S.I.C., Madrid, 105 pp. antunes, m. t., azanza, b., morales, j. and soria, d. 1994. Los Rumiantes (Artiodactyla. Mammalia) del Mioceno inferior del area de Lisboa (Cuenca del Tajo, Portugal). Revista Espahola de Paleontologia , 9, 51-71. azanza, b. 1993a. Systematique et evolution du genre Procervulus , cervide (Artiodactyla, Mammalia) du Miocene inferieur d’Europe. Comptes Rendus de I'Academie des Sciences de Paris , 316, 717-723. 19936. Sur la nature des appendices frontaux des cervides (Artiodactyla, Mammalia) du Miocene inferieur et moyen. Remarques sur leur systematique et leur phylogenie. Comptes Rendus de I'Academie des Sciences de Paris , 316, 1163-1169. - and menendez, E. 1990. Los Ciervos fosiles del Neogeno espanol. Paleontologia i Evolucid , 23, 75-82. beliajeva, e. i. 1974. On the history of the Asiatic Tertiary muntiacines. In kramarenko, n. n. (ed ). [Mesozoic and Cenozoic faunas and biostratigraphy of Mongolia.] Transactions of the Joint Soviet- Mongolian Paleontological Expedition. 1. 80-86. [In Russian]. bohlin, B. 1937. Eine tertiare Saugetierfauna aus Tsaidam. Palaeontologia Sinica , 15, I 111. bruijn, h. de, daams, r., daxner-hock, g., fahlbusch, v., ginsburg, l., mein, p. and morales, j. 1992. Report of the RCMNS working group on fossil mammals, Reisensburg 1990. Newsletters on Stratigraphy , 26. 65-1 18. bubenik, a. b. 1975. Taxonomic value of antlers in Genus Rangifer. 41-63. In smith, h., luick, j. r., lent, p. c., klein, d. r. and white, r. G. (eds). Proceedings of the First International Reindeer and Caribou Symposium, Alaska, 1972. Biological Papers of the University of Alaska. Special Report 1. 1983. Taxonomy of Pecora in relation to morphology of their cranial appendices. 163-185. In brown, r. d. (ed.). Antler development in Cervidae. Caesar Kleberg Wildlife Research Institute, Kingsville (Texas), 480 pp. 1990. Epigenetical, morphological, physiological, and behavioral aspects of evolution of horns, pronghorns, and antlers. 3-113. In bubenik, g. a. and bubenik, a. b. (eds). Horns , pronghorns and antlers. Springer-Verlag, New York, 562 pp. bubenik, g. a. and larsen, d. G. 1990. Velericorn antlers on a mature male moose (Alces a. gigas). Alces , 26, I 15-128. bubenik, g. a. 1990. Neuroendocrine regulation of the antler cycle. 265-297. In bubenik, g. a. and bubenik, a. B. (eds). Horns , pronghorns and antlers. Springer-Verlag, New York, 562 pp. and bubenik, a. b. 1986. Phylogenie and ontogenic development of antlers and neuroendocrine regulation of the antler cycle - a review. Saugetierkundliche Mitteilungen , 33, 97-123. chapman, d. i. 1975. Antlers, bones of contention. Mammal Review , 5, 121 172. chow ben-shun and shih ho-chunang. 1978. A skull of Lagomeryx from the middle Miocene of Linchu, Shangtung. Vertebrata PalAsiatica , 16. I I 1-122. Colbert, E. H. 1936. Tertiary deer discovered by the American Museum asiatic expeditions. American Museum Novitates, 854, 1-21. cook, H. j. 1934. New artiodactyls from the Oligocene and Lower Miocene of Nebraska. The American Midland Naturalist, 15, 148-165. CRUSAFONT, m. 1952. Los Jirafidos fosiles de Espana. Memorias y Comunicaciones del Institute Geoldgico, Diputacion Provincial de Barcelona, 8, 239 pp. DF.NizoT, G. 1927. Les formations continentales de la region orleanaise. Launay et Fils, Vendome, 582 pp. 484 PALAEONTOLOGY, VOLUME 40 eisenberg, J. F. 1987. The evolutionary history of the Cervidae with special reference to the South American radiation, 60-64. In wemmer, l. (ed.). Biology and management of the Cervidae. Smithsonian Institution Press, Washington, 577 pp. frick, c 1937. Horned ruminants of North America. Bulletin of the American Museum of Natural History, 69, 1-669. gabounia, l. k. 1959. [On the fossil muntjaks from the middle Miocene of the Caucasus.] Paleontologicheskii Zhurnal, 1, 114-117. [In Russian]. 1973. [Fossil vertebrate fauna of Belometsclieskaya.] Metsniereba, Tibilisi, 136 pp., [In Russian], gaudry, A. 1878. Les enchainements du monde animal dans les temps geologiques : Mammiferes tertiaires. Masson, Paris, 295 pp. gentry, a. w. 1994. The Miocene differentiation of old world Pecora (Mammalia). Historical Biology , 7, 115-158. — and heizmann, e. p. j. 1993. Lagomeryx Roger, 1904 (Mammalia, Artiodactyla): proposed designation of L. ruetimeyeri Thenius, 1948 as the type species. Bulletin of Zoological Nomenclature, 50, 133-136. ginsburg, l. 1972. Sur Page des mammiferes des faluns miocenes du Nord de la Loire. Comptes Rendus de I'Academie des Sciences de Paris, 274, 3345-3347. 1985. Essai de phylogenie des Eupecora (Ruminantia, Artiodactyla, Mammalia). Comptes Rendus de I'Academie des Sciences de Paris, 301, 1255-1257. 1989. Les mammiferes des sables du Miocene inferieur des Beilleaux a Savigne-sur-Lathan (Indre-et- Loire). Bulletin du Museum national dHistoire Naturelle Paris, 11, 101-121. — 1990. The faunas and stratigraphical subdivisions of the Orleanian in the Loire Basin (France). 157-176. In lindsay, E. H., FAHLBUSCH, V. and MEIN, p. (eds). European Neogene mammal chronology. Plenum Press, New York, 658 pp. — and azanza, b. 1991. Presence de bois chez les femelles du cervide nriocene Dicrocerus elegans et remarques sur le probleme de l’origine du dimorfisme sexuel sur les appendices frontaux des cervides. Comptes Rendus de I'Academie de Sciences de Paris, Serie 2, 313, 121-126. huin, j. and locher, j. p. 1985. Les Artiodactyles selenodontes du Miocene inferieur des Beilleaux a Savigne-sur-Lathan (Indre-et-Loire). Bulletin du Museum national dHistoire Naturelle, Paris, 7, 285-303. — and ukkakimapan, j. 1983. Un Cervide nouveau du Miocene d’Asie et Page des lignites des bassins intramontagneux du Nord de la Thailande. Comptes Rendus de I'Academie des Sciences de Paris, 297, 297-300. GOSS, R. J. 1970. Problems of antlerogenesis., Clinical Orthopaedics and Related Research, 69, 227-238. 1983. Deer Antlers. Regeneration, function and evolution. Academic Press, New York, 316 pp. Hamilton, w. r. 1978. Cervidae and Palaeomerycidae. 540-572. In maglio, v. j. and cooke, h. b. s. (eds). Evolution of African mammals. Harvard University Press, Cambridge, 641 pp. heintz, e. 1970. Les cervides villafranchiens de France et d’Espagne. Memories du Museum National d'Histoire Naturelle, Paris, Serie C, 22, 303 + 206 pp. janis, c. m. and scott, k. m. 1987. The interrelationships of higher ruminant families with special emphasis on the members of the Cervoidea. American Museum Novitates, 2893, 1-85. leinders, j. j. m. 1983. Hoplitomerycidae fam. nov. (Ruminantia, Mammalia) from Neogene fissure fillings in Gargano (Italy). Part 1, the cranial osteology of Hoplitomeryx gen. nov. and a discussion on the classification of pecoran families. Scripta Geologica , 70, 1-68. — and heintz, e. 1980. The configuration of the lacrimal orifices in pecorans and tragulids (Artiodactyla, Mammalia) and its significance for the distinction between Bovidae and Cervidae. Beaufortia, 30, 175-162. LINNAEUS, c. 1758. Systema naturae per regna tria naturae, secundum classes, ordines, genera, species cum characteribus, differentiis, synonymis, locis. Laurentii Salvii, Stockholm, vol. I, 824 pp. MacINNES, d. 1936. A new genus of fossil deer from the Miocene of Africa. Linnean Society Journal of Zoology, 34, 521-530. mayet, l. 1908. Etude des Mammiferes miocenes des sables de POrleanais et des Faluns de la Touraine. Annates de I'Universite de Lyon , 24, 336 pp. morales, j., pickford. m. and soria, d. 1994. Pachyostosis in a Lower Miocene giraffoid from Spain, Lorancameryx pachyostoticus nov. gen. nov. sp., and its bearing on the evolution of bony appendages in artiodactyls. Geobios, 26, 207-230. Owen, r. 1848. Description of teeth and portions of jaws of two extinct anthracotherioid quadrupeds discovered in the Eocene deposits on the NW coast of the Isle of Wight. Quarterly Journal of the Geological Society, London, 4, 103-141. pilgrim, G. e. 1941. The dispersal of the Artiodactyla. Biological Review, 16, 134—163. AZANZA AND GINSBURG: LAGOMERYCID ARTIODACTYLS 485 Roger, o. 1904. Wirbeltierreste aus dem Obermiocan der bayerisch-schwabischen Hochebene. Berichte des Naturwissenschaftlichen Vereins fur Schawaben und Neuburg (a.V.) Ausburg , 36, 3-22. Rothschild, m. and neuville, h. 1910. Recherches sur l’Okapi et les Giraffes de l’Est africain. Annales des Sciences Nature lies , Zoologie, 10, 1-93. rutimeyer, l. 1880. Beitrage zu einer natiirlichen Geschichte der Hirsche. Abhandlungen der Schweizerischen Palaontologische Gessellschaft , 7, 1-8. schlosser, m. 1904. Notizen iiber einige Saugetierfaunen aus dem Miozan von Wiirttemberg und Bayern. Neues Jahrbuch fur Mineralogie, Geologie und Palaontologie , 19, 485-502. SCOPOLi, G. A. 1 111 . Introductio ad historian i naturalem , sistens genera lapidum , plantarum et animalium hactenus detecta , caracteribus essentialibus donata , in tribus divisa , subinde ad leges naturae. Gerle, Prague, 506 + 34 pp. SIMPSON, G. G. 1931 . A new classification of mammals. Bulletin of the American Museum of Natural History , 59, 259-293. — 1945. Principles of classification and a classification of mammals. Bulletin of the American Museum of Natural History. 85, 1-350. solounias, N. 1988. Evidence from horn morphology in the phylogenetic relationships of the pronghorn ( Antilocapra americana). Journal of Mammalogy, 69, 140-143. stehlin, H. G. 1907. Noticies paleomammalogiques sur quelques depots miocenes des bassins de la Loire et de l’Allier. Bulletin de la Societe Geologique de France , 7, 525-550. — 1937. Bemerkungen iiber die miocanen Elirschgenera Stephanocemas und Lagomeryx. V erhandlungen der Naturforschenden Gesellschaft in Basel, 48, 193-214. 1939. Dicroceros elegans Lartet und sein Geweihwechsel. Eclogue Geologica Helvetiae, 32, 162-179. steininger, f. f., rabeder, G. and rogl, f. 1985. Land mammal distribution in the Mediterranean Neogene: a consequence of geokinematic and climatic events. 559-571. In Stanley, d. j. and wezel, f. c. (eds). Geological evolution of the Mediterranean Basin. Raimondo Sel/i Commemoration Volume. Springer Verlag, New York, 589 pp. TEDFORD, R. H., SKINNER, M. F., FIELD, R. W., ROMSBERGER, J. M., WHISTLER, D. P., GALUSHA, T., TAYLOR, B. E., Macdonald, j. r. and webb, s. d. 1987. Faunal succession and biochronology of the Arikareean through Hemphillian interval (Late Oligocene through earliest Pliocene epochs) in North America. 151-210. In woodburne, m. o. (ed.). Cenozoic mammals of North America. Geochronology and biostratigraphy. University of California Press, Berkeley, 336 pp. thenius, e. 1948. Uber ein stammesgeschichtlich interessantes Stadium aus der Geschichte der Hirsche. Anzeiger der Osterreichischen Akademie der Wissenschaften. Mathematisch-Naturwissenschaftliche Klasse , 14, 219-224. vislobokova, I. a. 1983. [Fossil deer of Mongolia.] Trudy Sovmestnaya Sovetsko-MongoT skaya Paleonto- logicheskaya Ekspeditsiya, 23, 1-75. [In Russian]. — hu changkang and sun bo 1989. On the systematic position of the Lagomerycinae. Vertebrata PalAsiatica, 27, 128-132. voorhies, m. r. 1969. Taphonomy and population dynamics of an Early Pliocene vertebrate fauna, Knox County, Nebraska. Contribution to Geology , University of Wyoming, Special Paper 1, 1-69. young chung chien 1937. On a Miocene Mammalian Fauna from Shantung. Bulletin of the Geological Society of China, 17, 209-244. — 1964. On a new Lagomeryx from Lantian, Shansi. Vertebrata PalAsiatica, 11, 329-340. BEATRIZ AZANZA Departamento de Paleobiologta Museo Nacional de Ciencias Naturales Jose Gutierrez Abascal, 2 28006-Madrid, Spain LEONARD GINSBURG Institut de Paleontologie Museum National d'Histoire Naturelle 8, rue Buffon, 75005-Paris, France Typescript received 29 September 1995 Revised typescript received 29 March 1996 A NEW TRANSITIONAL MYALINID BIVALVE FROM THE LOWER PERMIAN OF WEST TEXAS by CHRISTOPHER A. MCROBERTS and NORMAN D. NEWELL Abstract. Novaculapermia boydi gen. et sp. nov. is a remarkable Lower Permian vertically elongate bivalve that superficially resembles ‘razor’ clams of the superfamily Solenoidea. Our 'razor' clam possessed a duplivincular ligament and early ontogeny of the Myalinidae. The flattened, equiconvex form suggests that Novaculapermia was a shallow vertical burrower in soft sediments possibly anchored by byssal attachment; evidently it was not a reef dweller, but lived in near-reef environments. The silicified fauna of the Permian Glass Mountains of west Texas has provided one of the world’s most spectacular palaeontological windows into Late Palaeozoic reef and associated environments. Although well known for its rich brachiopod fauna (e.g. Grant 1971; Cooper and Grant 1972), other groups, such as the sponges (e g. Finks 1960) and bivalves (e.g. Newell and Boyd 1970, 1995), are numerically less diverse, but an equally important constituent of the west Texas Lower Permian fauna. Together with the Permian reefs of China and Tunisia, the Texas reefs testify to the morphological and ecological complexity achieved in Late Palaeozoic tropical marine biotas. Among tropical Permian Bivalvia, the Myalinidae and Alatoconchidae perhaps best illustrate morphological specialization and adaptation to a variety of reef and level-bottom habitats. The myalinids are a diverse group of marine and non-marine bivalves extending back at least to the early Carboniferous and reaching their developmental zenith during the mid Permian. They are an especially noteworthy group because several independent lineages within the family provide numerous classic examples of ‘progressive’ evolutionary trends related to their transition from an endobyssate to epibyssate life-habit (Newell 1942; Stanley 1972). Equally striking are the Alatoconchidae, the giant clams of the Permian, whose extreme size and large wing-like medial projections enabled these bivalves to recline on soft substrates without the aid of a strong byssus. The close affinities between the Alatoconchidae and Myalinidae were known soon after their first description and in fact they were grouped in a single family by Runnegar and Gobbett (1975). Differences in the ligament and valve symmetry have subsequently required assignment of these giant clams into the Alatoconchidae, a family based on poorly preserved material from Afghanistan (Ternner et al. 1973), which are presumably derived from the myalinids (Yancey and Boyd 1983). We describe here a collection of bivalves whose form is unlike that of any other known to us and which show characteristics of both the Myalinidae and Alatoconchidae. The specimens were collected from near-reef sedimentary rocks of the Lower Permian Road Canyon and Cathedral Mountain formations of the Glass Mountains of west Texas. A more complete account of the Permian stratigraphy and geology of the Glass Mountains and neighbouring Guadalupe Mountains can be found in Cooper and Grant (1972) and Newell et al. (1972). The specimens are preserved as silicified pseudomorphs in which fine skeletal detail can be discerned, although the original shell ultrastructure has been obscured during diagenesis. The significance of the shells is that they exhibit morphological features which, although known independently from unrelated bivalve clades, have never been described in combination. They represent yet another example of extraordinary morphological development among the fauna of the Permian Glass Mountains and provide clues to the evolutionary relationships among Late Palaeozoic bivalves. | Palaeontology, Vol. 40, Part 2, 1997, pp. 487-495, 1 pl.| © The Palaeontological Association 488 PALAEONTOLOGY, VOLUME 40 SYSTEMATIC PALAEONTOLOGY Institutional abbreviations: AMNH, American Museum of Natural History (New York); USNM, Museum of Natural History Smithsonian Institution (Washington). Following the recommendation by the International Code of Zoological Nomenclature (Ride et at. 1985), the suffix ‘-oidea’ is used in superfamily names. Order pterioida Newell, 1965 Superfamily ambonychioidea Miller, 1877 Family myalinidae Freeh, 1891 Genus novaculapermia gen. nov. Type species. Novaculapermia boydi sp. nov., by monotypy. Derivation of name. From a combination of Latin novacula, in reference to its razor-like shape, and permia, in reference to its only known occurrence from Permian strata. Diagnosis. Equivalved Myalinidae; shell large, showing change from retrocrescent to infracrescent during growth, adults adapically elongated; possessing simple opisthodetic duplivincular ligament, ligamental grooves slightly curved, intersecting hinge margin; valve width narrow; pallial line deeply bilobed ; probably anisomyarian, posterior adductor scar large, sub-circular, positioned in dorsal lobe of pallial impression; surface sculptured with commarginal growth squamae, lacking radial sculpture. Remarks. This genus is placed tentatively within the Myalinidae because of its possession of an opisthodetic duplivincular ligament. It differs from other myalinids in possessing an umbonal groove, in its narrow valve width and equivalved condition. New material, documenting the shell structure, however, may warrant erection of a new family to accommodate the unusual morphology. Novaculapermia boydi sp. nov. Plate 1 ; Text-figure 1 Derivation of name. After Donald C. Boyd, significant contributor to our understanding of Permian Bivalvia. Holotype. USNM 487771 ; from USNM loc. 702. Paratypes. USNM 487772^487773; from AMNH loc. 500 x and USNM loc. 702-un respectively. Material. The new species is based on six nearly complete valves representing four individuals, in addition to abundant fragmentary material. Localities and age. The specimens come from the Road Canyon and Cathedra) Mountain formations, west Texas, which are considered to be Middle Permian (Leonardian) in age (Newell et al. 1953). Complete EXPLANATION OF PLATE 1 Figs 1-8. Novaculapermia boydi gen. et sp. nov.; Cathedral Mountain Formation, Glass Mountains, Texas. 1-4, USNM 487772, paratype; adult individual; AMNH loc. 500 x . I, right valve exterior. 2, right valve interior showing well defined bilobate pallial line. 3, left valve interior. 4, left valve exterior. Note that the exterior surface of both valves is encrusted indicating probable excavation and exposure above the sediment- water interface. 5-8, USNM 487773, paratype; adult individual; USNM loc. 702-un. 5, right valve exterior. 6, right valve interior. 7, left valve interior. 8. left valve exterior showing epibiont encrustation. All x 0 75. PLATE 1 iliiwB” BKkisMpBgs m m I | &|S® WM'Z ? ?*- v ' £v^ ifg vK^i McROBERTS and NEWELL, Novaculapermia 490 PALAEONTOLOGY, VOLUME 40 text-fig. 1 . Novaculapermia boydi gen. et sp. nov. ; holotype, USNM 487771 ; Cathedral Mountain Formation, Glass Mountains Texas (USNM loc. 702). a, left valve interior showing broad flattened opisthodetic duplivincular ligamental area with eight acute ligamental grooves and shallow excavated umbonal groove. b, exterior of same specimen showing commarginal growth lamellae. Bothx 1-5. descriptions of localities (AMNH loc. 500 x , USNM Iocs. 702, 702-un, 703-c, 706-c) can be found in Cooper and Grant (1972). Diagnosis. As for genus. Description. A schematic diagram of the internal morphological features and shell measurements is provided in Text-figure 2. An orientation similar to that proposed for the myalinids and alatoconchids, with the apical end anterior and adapical end posterior (Newell 1942; Runnegar and Gobbett 1975), is favoured, even though the inferred living position does not reflect this orientation. The shells of Novaculapermia are fairly large, often exceeding 130 mm in their greatest curvilinear dimension measured along the vector of maximum growth. The rounded umbones occupy the anteriormost extremity of the shell which, during early growth stages, is retrocrescent and later becomes infracrescent. The exterior surface of both valves is sculptured with fairly coarse and unevenly spaced commarginal growth lamellae. Both the strength of, and spacing between individual lamellae is apparently equal for the two valves of the same individual. The valves are relatively flat, lacking an umbonal ridge or keel. The maximum width of two valves when placed in opposition is very narrow (< 10 mm). There is a slight gape (1-2 mm wide) at the posterior adapical end in one specimen with complete margins (PI. 1, figs 1-2). Although there is no evidence of a byssal collar or byssal gape along the antero-ventral margin in adult specimens, there appears to be a slight infolding of the shell margin in this position which might indicate passage of a byssus. A more pronounced byssal gape may have existed during early life stages, yet the accretion of new shell during growth has obscured the relationship between the two valves along the juvenile commissure margin precluding any such observation. Muscle scars can be observed on the valve interior of a few well preserved specimens. The most pronounced features observed on the shell interior are the bilobate pallial muscle insertion scars. These scars, observed in both right and left valves (PI. 1, figs 2-3, 6-7), are bilobate; the dorsal lobe is the larger, extending beyond half the length of the shell. The ventral lobe is narrow and abuts against the antero-ventral shell margin. A sharp medial constriction of the pallial line separates the two lobes and is presumably formed by the fixation of the pallial muscle. The pallial scar may have been diagenetically enhanced by the subsequent removal of nacreous shell enclosed by the pallial line prior to silicification. Situated within the dorsal pallial lobe at about half its distance is a faint impression of a large (c. 9 mm in diameter) sub-circular muscle insertion scar, presumably indicating the position of the posterior adductor muscle. Evidence for additional musculature has not been McROBERTS AND NEWELL: PERMIAN BIVALVE 491 text-fig. 2. Morphological features and orien- tation for a right valve interior of Novacula- permia. Note: orientation is similar to that used by Runnegar and Gobbett (1975) and Yancey and Boyd (1983) for the Alatoconchidae. umbonal groove DISTANCE FROM UMBO (mm) text-fig. 3. Post-larval growth sequence of Novaculapermia showing the retrocrescent to infracrescent trend in ontogeny, a. = the angle between the tangent to the vector of maximum growth at the valve margin and the hinge axis, a, graphical illustration of growth series in the holotype (Text-fig. 1) where the horizontal line approximates the hinge axis. Scale bar represents 10 mm. b, plot showing change in a at a given distance from the umbo in three specimens. recognized. Both the shape and position of the pallial line and posterior adductor scar are strikingly similar to those found in certain pinnaceans, such as Pinna and Exitopinna, where the inner shell layer of the area enclosed by the pallial muscles is nacreous (see Cox and Hertlein 1969, fig. C23.3). These similarities in the shape of musculature among elongated semi-infaunal bivalves are believed to reflect a constructional constraint in the position and size of mantle cavity due to shell growth. Extending from the umbone along an arc parallel to shell growth and extending into the body cavity is a shallow U-shaped groove which is difficult to interpret. The groove has only been observed in a single well preserved left valve (Text-fig. 1). The groove is about 4 mm wide and exhibits convex (towards the beak) growth lines. On one hand, this structure may serve as an umbonal deck with a function similar to that of the umbonal septum described from other myalinids (Newell 1942) and some alatoconchids (e.g. Boyd and Newell 492 PALAEONTOLOGY, VOLUME 40 1979, fig. 9). If this were the case the groove might reflect the migration tract of the anterior adductor muscles during growth. On the other hand, the groove may have had a function similar to that of the byssal groove of some other alatoconchids (e.g. Yancey and Boyd 1983) and represent the passage for a byssus. The first interpretation is favoured here because it is unlikely that a byssus would be channelled in an orientation unknown among other byssate bivalves. The ligamental characters of Novaculapermia can be deduced from the valve interior of one well preserved left valve (Text-fig. 1). Right valves showing an equivalent ligamental area are not preserved, but presumably would show a similar form. In this specimen, the broad flat ligamental area lies entirely behind the beak and is bounded antero-ventrally by the shallow umbonal groove. At least eight thin grooves, in which lamellar bands of an external ligament were inserted, transverse across the ligamental area. The spacing between the grooves (c. 1 mm) is several times greater than the width of the grooves. The ligament insertion grooves are only slightly curved and are inclined towards the hinge axis. Two of the grooves intersect the hinge axis at an angle of about 10°. Unfortunately, articulated valves with an intact ligament are unknown, but, provided our assessment of ancestry is correct, Novaculapermia probably had a ligament similar to that of Septimyalina , described by Carter (1990), in which the fibrous sublayers extended continuously from one valve to the other. Growth sequences of Novaculapermia illustrate a post-larval ontogenetic trend in shell shape from retrocrescent to infracrescent (Text-fig. 3). During the retrocrescent juvenile stage (up to about 30 mm long), the growth vector is curved anteriorly. After about 30 mm the growth vector ceases to be curved and the adult shape becomes infracrescent and continues accreting new shell along a straight vector nearly normal to the hinge axis. Similar ontogenetic trends have been identified in several myalinid species, and are thought to have arisen independently in several different lineages (Newell 1942). Such radical changes in growth have traditionally been interpreted as reflecting a change in living habits, and among several myalinid lineages similar trends probably reflect the transition from a semi-infaunal to an epifaunal habit (Stanley 1972). A new problem arises in that Novaculapermia was probably not an epifaunal bivalve (see below) and such an interpretation for the trend is not warranted in this case. Remarks and comparisons. This species is unlike any known to us and it is unlikely that it would be confused with other Permian Bivalvia. Given the few samples available to us, it is impossible to know the limits of variation within the species. Some variation does exist in shell shape and observable features (e.g. ligamental area and musculature) which may be due to preservational factors, genetic differences or ecophenotypic variation. The discovery and description of additional material of a sufficient sample size may require the separation of two or more distinct species. PALAEOECOLOGY Like other pteriomorphs, we assume that Novaculapermia was a filter feeder utilizing a suspended food source. Because the specimens were not recovered in situ , the life orientation for Novaculapermia bovdi remains questionable and can only be inferred using a functional morphological approach. Such dependence on purely functional morphology in reconstructing life- habits does have its pitfalls (see discussion in Fursich 1980), yet remains a powerful, and sometimes the only, tool in the absence of unequivocal field evidence or modern analogues. The following discussion is considered only preliminary; we await further evidence from an in situ association. Several clues indicate that Novaculapermia boydi was sessile, semi-infaunal and oriented with its sagittal plane normal to the sediment-water interface. A hypothesized reconstruction is provided in Text-figure 4 and shows a habit similar to that proposed for other 'mud sticking’ bivalves such as Cochlearites and Lithiotis (e.g. Seilacher 1984). The equivalved condition is an ubiquitous feature in bivalves which have their sagittal plane oriented vertically (e.g. Stanley 1970, 1972), a condition which can also be inferred for Novaculapermia. If the interpretation of the anterior marginal fold as a passage for a byssus is correct, then Novaculapermia may have employed a byssus during its entire life, which would have aided in fixation, especially in more agitated water where sediments would be more likely to shift. We interpret Novaculapermia to be only weakly byssate given the absence of a distinct byssal gape. Byssal attachment is not a prerequisite for a semi-infaunal habit because Novaculapermia could also have been supported by enclosing sediment, which would also have aided in stabilization, especially in less agitated water. Although the question of how deeply McROBERTS AND NEWELL: PERMIAN BIVALVE 493 text-fig. 4. Reconstruction of hypothesized life habit for Novaculapermia. Arrows indicate relative position of inhalant and exhalant currents. Novaculapermia was buried in the sediment is difficult to assess, a clue can be found on the two valves of one specimen (PI. 1, figs 5-8) where the posterior margins are broken 10-20 mm from their former posterior shell margin. Such breakage, frequently induced by predators, is common amongst semi-infaunal pinnids such as Pinna carnea which are typically buried to a depth of about half their greatest dimension (Stanley 1970). Although such a shallow burial depth may be possible, we favour a slightly deeper burial depth on the grounds that it would be more stable for a weakly byssate Novaculapermia. M YAL1NID AFFINITIES The specimens referred to herein exhibit many morphological features which suggest affinities with several different bivalve groups depending on which morphological characters are emphasized. Although a duplivincular ligament is believed to have been derived independently among several bivalve clades (Newell and Boyd 1987) it is a consistent feature shared among all Myalinidae. The ligament system observed in Novaculapermia is perhaps most similar to that found in the myalinids Liebea and Septimyalina , in which the ligament grooves are fewer in number and intersect the hinge line at a steep angle. This is in contrast to the ligament of Selenimyalina and some Myalina sensu stricto where the numerous fine ligament grooves incise a narrow ligament area and run nearly parallel to the hinge axis. It is also possible to argue for homology in the umbonal features (umbonal deck and groove) between myalinids and Novaculapermia depending upon their interpretation. We favour an interpretation of the umbonal groove and deck as homologous to the umbonal deck of some myalinids. An additional similarity of Novaculapermia to the Myalinidae is that of general shell shape, especially early in ontogeny. In both Novaculapermia and many Upper Carboniferous and Permian myalinids, shell shape changes from markedly retrocrescent mytiliform to an upright infracrescent position. The genus Novaculapermia also exhibits some obvious similarities to members of the Alatoconchidae which are thought to have been derived from the Myalinidae (Runnegar and Gobbett 1975; Yancey and Boyd 1983; Yancey and Ozaki 1986). Like the myalinids, alatoconchids also have a duplivincular ligament and a broad ligament area. Unlike the myalinids and Novaculapermia , the ligamental grooves of alatoconchids are somewhat sinuous and not straight or slightly arched. Like Novaculapermia , alatoconchids are equivalved; a condition unknown in other myalinids. However, Novaculapermia lacks the pronounced alate wing-like flanges, byssal collar, and extreme large size characteristic of the alatoconchids. We concluded that Novaculapermia shares more common features with the Myalinidae than the Alatoconchidae and prefer to place it in the former. We believe that the similarities in the two 494 PALAEONTOLOGY, VOLUME 40 groups stem from their recent common ancestry. A more complete treatment of phylogenetic affinities among the myalinids and their ancestors is underway (McRoberts and Newell 1995). Acknowledgements. This work would not have been possible without the diligent collecting and preparation of Glass Mountains material by G. A. Cooper who kindly made available the specimens described herein. Thanks also to M. Florence for assisting with curation of the specimens into the Smithsonian and H. Schirm for some of the photography. J. Scallan, F. Fiirsich, and M. Aberhan and an anonymous reviewer made useful suggestions leading to a better manuscript. Much of this work was conducted while CAM was an Alexander von Humboldt-Stiftung Postdoctoral Fellow; the foundation’s support is gratefully acknowledged. REFERENCES boyd, d. w. and Newell, N. d 1979. Permian pelecypods from Tunisia. American Museum Novitiates, 2686, 1-22. carter, j. G. 1990. Evolutionary significance of shell structure in the Palaeotaxodonta, Pteriomorphia, and Isofilibranchia (Bivalvia; Mollusca). 135-295. In carter, j. g. (ed.). Skeletal biomineralization, Volume 1 Von Nostrand Reinhold, New York. cooper, G. a. and grant, r. e. 1972. Permian brachiopods of west Texas. E Smithsonian Contributions to Paleobiology, 14, 1-183. cox, l. r. and hertlein, L. G. 1969. Superfamily Pinnacea. N281-N285. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part N. Mollusca 6, Bivalvia 1 . Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. finks, R. m. 1960. Later Paleozoic sponge faunas of the Texas region. Bulletin of the American Museum of Natural History , 120. 1-160. frech, f. 1891. Die devonischen Aviculiden Deutschlands. Abhandlungen zu den geologischen Specialkarte von Preussen und den Thiiringischen Staaten, 9, 1-253. fursich, f. t. 1980. Preserved life positions of some Jurassic bivalves. Palaontologische Zeitschrift, 54, 289-300. grant, r. E. 1971. Brachiopods in the Permian reef environment of west Texas. Proceedings of the North American Paleontological Convention , Part J, 1444-1481. McRoberts, c. a. and Newell, N. D. 1995. Phylogeny and cladistic clutter of Paleozoic and Early Mesozoic marine Myalinidae. 21. In johnston, p. a. (ed.). The fifth Canadian paleontology conference and international symposium on the paleobiology and evolution of the Bivalvia. Geological Association of Canada, Drumheller. miller, s. a. 1877. The American Palaeozoic fossils. Published by the author, Cincinnati, 245 pp. newell, n. d. 1942. Late Paleozoic pelecypods: Mytilacea. Publication of the Kansas State Geological Survey, 10, 1-115. 1965. Classification of the Bivalvia. American Museum Novitiates, 2206, 1—25. — and boyd, d. w. 1970. Oyster-like Permian Bivalvia. Bulletin of the American Museum of Natural History, 143, 217-282. — 1987. Iteration of ligament structures in pteriomorphian bivalves. American Museum Novitiates, 2875, 1-11. 1995. Pectinoid bivalves of the Permian-Triassic crisis. Bulletin of American Museum of Natural History, 227 , 1-95. — rigby, j. k., fiscuer, a. G., hickox, J. E. and Bradley, j. s. 1972. The Permian reef complex of the Guadalupe Mountains region, Texas and New Mexico. Hafner Publishing Company, New York, 236 pp. ride, w. d. l., sabrosky, c. w., bernardi, g. and melville, r. v. (eds). 1985. International code of zoological nomenclature. Third edition. International Trust for Zoological Nomenclature, London, 338 pp. runnegar, b. and gobbett, d. 1975. Tanchintongia gen. nov., a bizarre Permian myalinid bivalve from West Malaysia and Japan. Palaeontology, 18, 315-322. seilacher, a. 1984. Constructional morphology of bivalves: evolutionary pathways in primary versus secondary soft-bottom dwellers. Palaeontology, 27, 207-237. Stanley, s. m. 1970. Relation of shell form to life habits in the Bivalvia. Memoir of the Geological Society of America , 125, 1-296. 1972. Functional morphology and evolution of byssally attached mollusks. Journal of Paleontology, 46, 163-212. McROBERTS AND NEWELL: PERMIAN BIVALVE 495 termier, h., termier, G. and lapparent, a. f. de 1973. Grands bivalves recifaux du Permien superieur de 1" Afghanistan central. Annates Societe Geologie du Nord , 93, 75-80. yancey, t. e. and boyd, d. w. 1983. Revision of the Alatoconchidae : a remarkable family of Permian bivalves. Palaeontology , 26, 497-520. — and ozaki, k. 1986. Redescription of the Genus Shikamaia, and clarification of the hinge characters of the family Alatoconchidae (Bivalvia). Journal of Paleontology, 60, I 16-125. CHRISTOPHER A. McROBERTS Department of Geology SUNY Cortland Cortland, New York 1 3045, USA NORMAN D. NEWELL Department of Fossil Invertebrates American Museum of Natural History Typescript received 20 December 1995 Central Park West and 79th Street Revised typescript received 22 July 1996 New York, New York 10024, USA EARLY JURASSIC BRACHIOPODS FROM GIBRALTAR, AND THEIR TETHYAN AFFINITIES by ellis f. owen and edward p. f. rose Abstract. A spiriferinid Liospiriferina rostrata , two rhynchonellids Gibbirhynchia correcta and Pontaltorhynchia schopeni gen. nov., a terebratulid Merophricus mediterranean and a zeilleriid Calpella aretusa gen. nov., constitute the first invertebrate fauna to be described systematically from the ‘Gibraltar Limestone’. This formation, a 600 m thick cyclic sequence of well-cemented peritidal dolomitic limestones, has also yielded stromatolitic and oncoidal algae, a stromatoporoid, and locally common but generically indeterminate gastropods and bivalves. The brachiopods are of early Lias (Sinemurian) age, based on comparisons with faunas largely from Morocco and Sicily, and those described from the upper Sinemurian of the Central Apennines, Italy. The Sinemurian age verified for a significant part of the Rock of Gibraltar, an isolated and partly overturned klippe, allows its correlation with other tectonically displaced carbonates in the Betic-Rif arc which borders the western extremity of the Mediterranean Sea. Elements of a similar brachiopod fauna ranging from Italy through Gibraltar to Morocco are associated with thick Liassic platform carbonates which characterized the southern continental margin of the Tethys until its widespread collapse, generally in the mid-late Liassic. Gibraltar is a peninsula, 1-6 km in natural maximum width, 5-2 km long, and some 6 km2 in land area, which juts south from Spain at the western entrance to the Mediterranean Sea (Text- fig. 1). The peninsula is dominated by the Rock, a 600 m thick sequence of limestones and dolomites, the North Face of which rises precipitously to over 400 m above the northern isthmus, and which continues laterally for 2-5 km as a sharply ridged crest before descending to the two successive plateaux which truncate it to the south. The carbonates of Gibraltar form an isolated outcrop, flanked by Quaternary screes and sands and surrounded by the sea except for the low-lying isthmus of Quaternary sands which joins the peninsula northwards to mainland Spain, a region essentially of Tertiary flysch sandstones. The closest comparable carbonates occur at Los Pastores, on the southern edge of Algeciras, 9 km to the west across the Bay of Gibraltar; near Manilva, inland from Estepona, 30 km to the north of La Linea; and at Gebel Musa in Morocco, 24 km south across the Strait of Gibraltar. Their relationships are controversial, since stratigraphical ages are imprecise, and overall the principal structural-stratigraphical units of this western Mediterranean region have been subject to considerable displacement, by major strike-slip faulting and/or thrusting due to nappe emplacement associated with African-European continent-continent collision during the late Cenozoic (Rose and Rosenbaum 1994a). Structurally, Gibraltar lies at the boundary of the external and internal zones (Iberian and Alboran domains) long distinguished in the Betic region of south-east Spain, and gives its name to the Gibraltar Arc which unites the Betics southwards with the Rif mountains of Morocco. A Jurassic age for the Gibraltar carbonates has long been established on the basis of fossil identifications recorded by Smith ( 1 846 : terebratulid brachiopod casts), Ansted ( 1 859 : an ammonite and some terebratulid brachiopods), Roemer (1864: brachiopods and a gastropod), Hochstetter (1866: casts of brachiopods and molluscs), Ramsay and Geikie (1878: a rhynchonellid brachiopod), Choffat (1892: a large gastropod plus terebratulid and rhynchonellid brachiopods), Dubar and Le Maitre (1935: terebratulid and spiriferid brachiopods, a spongiomorphid, and gastropods), and Dubar (1942: brachiopods); this age was duly accepted by Reed (1949). Bailey (1952) listed [Palaeontology, Vol. 40, Part 2, 1997, pp. 497-513, 1 pi.) © The Palaeontological Association 498 PALAEONTOLOGY, VOLUME 40 identifications by H. M. Muir-Wood of six brachiopod species/subspecies based on the 26 specimens collected and mentioned by Smith (1846), with the inference that these indicated an Early Jurassic age (equivalent to the Lower Lias of north-west Europe). He also listed identifications by L. E. Spath of ammonites collected in 1943 by A. L. Greig from shales at the base of the North Face and east coast of the Rock. These specimens were small, immature or badly preserved, but the fauna was recognized as Domerian (Mid Lias) in age, and therefore provided evidence that the stratigraphical sequence in the main ridge of the Rock had been inverted. Arkell (1956) followed Bailey (1952) in ascribing the Gibraltar succession largely to the Lower Jurassic (Lias). However, on a more recent Spanish map (Fontbote 1970) it is depicted as Middle-Upper Jurassic. The total Jurassic invertebrate fauna recorded from Gibraltar is sparse, with component taxa hitherto either imprecisely or only tentatively identified. Only two species identifications have been based on rigorous systematic description and discussion: a brachiopod, Zeilleria cf. tenuiplicata Dubar (in Dubar 1942), and the spongiomorphid Stromatomorpha liasica Le Maitre (in Dubar and Le Maitre 1935). Few specimens have yet been illustrated. A comprehensive description and review is therefore given here to clarify age assignments of lithostratigraphical units recently distinguished and mapped on the Rock (Rose and Rosenbaum 1990, 1991a, 19916; Rosenbaum and Rose 1991). GEOLOGY OF GIBRALTAR Structurally, the Rock of Gibraltar is divisible into two components separated by a north- west-south-east fault, the ‘Great Main Fault' of Ramsay and Geikie (1878), which extends south- east from the Gibraltar docks as a 200 m wide zone of fractured bedrock. North of this fault, in the main ridge area of the Rock, the sequence is overturned and dips at moderate to high angles to the west. South of the fault, in the southern plateaux, the sequence is the ‘right way up’ and dips at moderate to high angles to the east. Correlation between the two components is based on lithology rather than biostratigraphy, but similarities in cyclic peritidal depositional conditions and subsequent dolomitization of the thick carbonate sequences which dominate both components are sufficiently close to indicate their broad equivalence. In consequence, it is possible to recognize a composite tripartite division of the Gibraltar bedrock succession as currently exposed. Little Bay Shale Formation Up to 15 m of red and green fissile mudstones with thin beds of fine sandstone and pebble conglomerate, together with thicker beds of dark grey dolomite, crop out at the base of cliffs in Little Bay on the west coast of the southern plateaux. No age-diagnostic fossils have been described from these rocks, but the beds dip steeply eastwards beneath uninverted Gibraltar Limestone so must represent the oldest exposed bedrock. Red and green fissile mudstones with thin chert bands can be traced as sheared exposures at the base of faulted limestone/dolomite cliffs northwards into Camp Bay. Farther north, vertical ‘shales’ of similar appearance were mapped by Ramsay and Geikie (1878) along the coast west of the main ridge of the Rock, but have largely been removed or obscured by construction works, so are known now from patchy exposures, boreholes for foundation works, or old records. The Dockyard Shales shown on Text-figure I have been designated (Rose and Rosenbaum 1990) on the basis of photographs made during construction of the dockyards at the beginning of the 20th century. They are known to have had an eastward dip contrasting with the steep westward dip of the carbonates close by in the main ridge of the Rock, but their relationship to the Little Bay Shales is unproven. Gibraltar Limestone Formation Most of the Rock is formed by a 600 m thick succession of massively bedded carbonates, passing gradationally from dark grey bituminous dolomites at the base through pale grey-white dolomites, then a distinctively well-bedded unit of dolomites and limestones, to a very thick, relatively OWEN AND ROSE: JURASSIC BRACHIOPODS 499 text-fig. 1. Map of bedrock (pre-Quaternary) geology of Gibraltar, indicating localities mentioned in the text (from Rose and Rosenbaum 1990: see original publication for cross sections drawn along lines indicated on this figure). 500 PALAEONTOLOGY, VOLUME 40 homogeneous sequence of light or medium grey fine-grained limestones, dolomitic at the base. Fenestral and peloidal mudstones and wackestones, stromatolitic algal laminites, and carbonate conglomerate lenses with scoured bases occur in shallowing and fining upward cycles of 1-3 m thickness (J. L. Wood, in Rose and Rosenbaum 19946). These are common features of tropical or sub-tropical tidal flats traversed by migrating channels. Fossils recorded from Gibraltar in the last century, as noted above, are inferred to have been collected largely from this formation. The sparse recorded fauna is partly a reflection of the difficulty of extracting specimens: the rock is strong and crystalline, very similar in general appearance to the Carboniferous Limestone of England and Wales. Where they do occur, fossils are almost invariably seen only in cross section. Moreover, since much of the lower (and most accessible) part of the sequence is formed by algal limestones and intraformational conglomerates of peritidal origin, its depositional conditions were largely unfavourable for the proliferation or preservation of shelly marine invertebrates. Catalan Bay Shale Formation Medium-bedded grey cherty limestones alternating with thinner beds of reddish-grey fissile mudstones are exposed beneath inverted Gibraltar Limestone near Catalan Bay; the south end of Sandy Bay; some 500 m farther south; and at Ammunition Jetty, all on the east coast. They are exposed also at the base of the North Face of the Rock. Quarries at Catalan Bay and the North Face yielded the ammonites to which Spath (in Bailey 1952) gave a Donrerian (Mid Lias) age. Early descriptions of Gibraltar (notably Ramsay and Geikie 1878) recognized only a single ‘shale’ formation, along the west coast, supposedly overlying the ‘limestone ’. By 1943 quarrying had exposed ‘shales’ also at the base of the North Face, and along the eastern coast. On the basis of similar mudstone lithology, red/green colour and association with cherts, Bailey (1952) also correlated all the ‘shales’ on Gibraltar as part of a single formation, but on tentative biostratigraphical evidence from the northern and eastern outcrops he inferred that the ‘shales’ underlie rather than overlie the ‘limestone’. However, since the western shales differ from the northern/eastern shales in aspects of their lithology, and their age equivalence is unproven, they were described and mapped as separate formations by Rose and Rosenbaum (1990, 1991a, 19916) and Rosenbaum and Rose (1991). The ammonites on which the Domerian (= late Pliensbachian) age of the Catalan Bay Shales is based ( Rhacophyllites Stella (J. de C. Sowerby), Rhacophyllites sp., Lytoceras aff. andax (Meneghini), Harpoceras sp., and Phylloceras aft'. Calais (Meneghini) from Catalan Bay quarry; and Lytoceras sp. from the North Face) cannot now be traced in the collections of the British Geological Survey, The Natural History Museum in London, the Gibraltar Museum, or elsewhere (despite advertisement in The Geological Curator and extensive search) so are presumed lost. Brachiopods are the only age-diagnostic fossils currently available. For specimens described below, NHM = Department of Palaeontology, The Natural History Museum, London; ONCP = Office National de Gestion des Collections Paleontologiques, Centre des Sciences de la Terre, Universite Claude Bernard, Lyon 1, Villeurbanne, France. SYSTEMATIC PALAEONTOLOGY Phylum brachiopoda Dumeril, 1806 Class articulata Huxley, 1869 Superfamily spiriferinacea Davidson, 1884 Family spiriferinidae Davidson, 1884 Genus liospiriferina Rouselle, 1977 Type species. Terebratulites rostratus Schlotheim, 1822, p. 260, pi. 16, fig. 4a-c, from the Lias of Balingen, Wiirttemberg, Germany. OWEN AND ROSE: JURASSIC BRACHIOPODS 501 Liospiriferina rostrata (Schlotheim, 1822) Text-figure 2a-c 1822 Terebratulites rostratus Schlotheim, p, 260, pi. 16, fig. 4a-c. 1832 Delthyris rostrata Schlotheim; Zieten, p. 50, pi. 35, fig. 1. 1871 Spirifer rostratus Schlotheim; Quenstedt, pi. 54, fig. 96. 1886 Spiriferina rostrata Schlotheim; Di Stefano, p. 35, pi. 1, figs 1-8. 1886 Spriferina haasi Di Stefano, p. 39, pi. 1, figs 9-10. 1977 Liospiriferina rostrata (Schlotheim); Rouselle, p. 164, pi. 1, figs 5a-c, 8a-c. Material. A single damaged ONCP specimen, EM20311, 35 0 mm long, 32-2 mm wide and 27-3 mm thick. text-fig. 2. A-c, Liospiriferina rostrata (Schlotheim, 1822); EM2031 1 ; dorsal (a), anterior (b) and lateral (c) views, showing rounded beak ridges, wide interarea and inflation of the dorsal umbo, d-f, Gibbirhynchia correcta (Di Stefano, 1886); EM20312; dorsal (d), anterior (e) and lateral (f) views. G-H, Pontaltorhynchia schopeni (Di Stefano, 1886); EM20313; dorsal (g) and lateral (h) views. ONCP specimens, inferred in the text to be from the Gibraltar Limestone at the north-east of the Rock; Lower Lias, Sinemurian; x 1-5. Description. Medium-sized, smooth spiriferinid. Ventral valve with massive umbo and sub-erect beak truncated by a large foramen; beak ridges rounded and smooth; lateral flanks of the valve steep with faint concentric growth lines. Anterior margin uniplicate, exposing a broad, shallow ventral sulcus which narrows posteriorly. Delthyrium triangular, bounded by extensive flat interareas. Dorsal valve posteriorly inflated, with less steep flanks than the ventral valve. A low dorsal fold extends anteriorly, complementing the shallow ventral sulcus at the anterior margin. 502 PALAEONTOLOGY, VOLUME 40 Remarks. The specimen described here was identified by E. de Verneuil as Spirifer tumidus (Roemer 1864; Hochstetter 1866), but without illustration or description. Ansted (1859) and Dubar and Le Maitre (1935) recorded spiriferids from Gibraltar, but without more precise identification. Rouselle (1977, p. 164) described the genus Liospiriferina to accommodate specimens from Mid Liassic localities in Morocco and north-east Spain. Terebratulites rostratus Schlotheim, originally described from the Lias of Germany, was designated type species. Examples of morphotypes identified as conspecific were listed (p. 165) and examples figured (pi. 1, figs 5-9). Of these, two specimens (figs 5, 8) are considered here to be typical of the specimen figured by Schlotheim (1822, pi. 16, fig. 4a-c) and to compare favourably with the Gibraltar specimen illustrated by our Text- figure 2a-c. Di Stefano (1886) described several spiriferinid species from the Lias of Sicily and the central Apennines, some of which were costate but many were smooth-shelled. Of these, Spiriferina haasi Di Stefano, 1886, although slightly larger, is proportionately similar to the specimen figured here from Gibraltar (Text-fig. 2a-c). Di Stefano gave a series of dimensions of four specimens collected from the Calcare grigio of Gullo near St. Antonio, Taormina, Sicily. Their mean length is calculated as 39 5 mm, width 39-0 mm and thickness 31-5 mm. The morphological features match the Gibraltar specimen closely and we have no difficulty in assigning Di Stefano’s Spiriferina haasi also to Liospiriferina rostrata (Schlotheim). Several other species of smooth shelled Spiriferinidae have been described from the Lower Lias of the central Apennines by Canavari (1884) and Di Stefano (1886), both of whom recorded Spiriferina rostrata (Schlotheim). A specimen which approaches those assigned to Spiriferina haasi Di Stefano in general morphological aspects was described as S. rostrata var. striata Schlotheim, by Canavari (1884, p. 77, pi. 9, fig. 2a-d), but this can be distinguished from Liospiriferina rostrata (Schlotheim) in having a more arcuate incurved beak, less inflated dorsal umbo and an ornament of faint striae covering the surface of both valves. NHM specimens BB 13837-1 3842, identified as Spiriferina serinensis (Gemmellaro, 1874) and said to have been collected from the Toarcian of Ait Daoud Ou Azzi, Morocco by the late R. V. Melville in 1952, can be compared to similar specimens BB2029 1-20292 collected from the davoei Zone ‘ Leptaena' Bed of the foreshore on the east side of Taormina Bay, Sicily by Dr M. K. Howarth in 1957. They differ from the specimen from Gibraltar, figured here, in their smaller dimensions, deeper ventral sulcus, shorter umbo and less inflated dorsal valve. There are no closer similarities with spiriferinid species listed by Dubar (1932) amongst Liassic brachiopod faunas of Morocco. EXPLANATION OF PLATE 1 Figs 1 -11. Merophricus mediterranea (Canavari, 1884). 1-3, NHM B85440; Gibraltar (inferred in the text to be from the Gibraltar Limestone; Lower Lias, Sinemurian); x L5. 4—6, NHM B 19524; showing the typical elongate-oval outline and equibiconvexity of the valves; Catalan Bay, Gibraltar, presumed Sinemurian; x 1-5. 7-9, NHM B85434; showing marginal plicae and almost smooth dorsal valve; same locality and horizon as figs 1-3; x 1 10-11, NHM BB 13872; Lower Lias, Ait-Oufella, High Atlas Mountains, Morocco; for comparison with specimens from Gibraltar; x 1-5. Figs 12-17. Pont alt orliynchia schopeni (Di Stefano, 1886), gen. nov. 12-14, NHM B 14969; Lower Lias, upper Sinemurian; Pontalto, central Apennines, Italy; showing rounded costae and typical broad oval anterior. 15-17, NHM B85427; Gibraltar, details as for figs 1-3; x F5. Figs 18-20. Gibbirhynchia correcta (Di Stefano, 1886); NHM B85428; Gibraltar, details as for figs 1-3. The subquadrate outline and costation are typical of the genus and are comparable to the specimens figured by Di Stefano (1886); x F5. Figs 21-23. Calpella aretusa (Di Stefano, 1886) gen. nov. NHM B85421; Gibraltar, details as for figs 1-3; showing somewhat less acutely biconvex valves but maintaining the same general outline as the specimen figured by Di Stefano (1886, pi. 2, fig. 54a-d); x F5. PLATE 1 OWEN and ROSE, Jurassic brachiopods 504 PALAEONTOLOGY, VOLUME 40 text-fig. 3. Ten transverse serial sections through the umbo of a specimen of Pont alt orhynchia schopeni (Di Stefano, 1886) gen. nov. showing the deep septa- lium, strong median dorsal septum and the ventrally deflected hinge plates. The numerals denote the distance in milli- metres between each section. NHM BB20290; Lower Lias, jamesoni Zone, Cape Taormina, Sicily. Scale bar repre- sents 5 mm. Superfamily rhynchonellacea Gray, 1848 Family rhynchonellidae Gray, 1848 Subfamily tetrarhynchiinae Ager, 1965 Genus gibbirhynchia Buckman, 1918 Type species. Gibbirhynchia gibbosa Buckman, 1918, p. 43, pi. 13, fig. 7, from the Middle Lias of South Petherton, Somerset. Gibbirhynchia correcta (Di Stefano, 1886) Plate 1, figures 18-20; Text-figure 2d-f 1886 Rhynchonella correcta Di Stefano, p. 65, pi. 2, figs 39-41. 1892 Rhynchonella correcta Di Stefano; Choffat, p. x. 1952 Rhynchonella correcta Di Stefano; Muir-Wood, in Bailey, p. 164. Material. Four NHM specimens from Gibraltar: B 85428 figured here (PI. 1, figs 18-20), with shell approximately 16-3mm long, 151 mm wide and 1 2-9 mm deep; B 85423, 16 5 mm long, 17-8 mm wide, 121 mm deep; B 85425, 16-4 mm long, 16-8 mm wide and 1 L9 mm deep; and B85424 (embedded in matrix). A single ONCP Gibraltar specimen. EM203 12, figured here as Text-figure 2d-f: 16 5 mm long, 141 mm wide and 13-3 mm deep. Diagnosis. Small, biconvex, semisphaeoridal, uniplicate rhynchonellid. Umbo short, beak slightly incurved, deltidial plates poorly exposed. Dorsal fold almost imperceptible; ventral valve with shallow sulcus and extensive linguiform extension. Description. Dorsal valve more acutely convex than the ventral valve, with a slight umbonal inflation and poorly developed fold which appears to develop just anterior to mid valve, ventral sulcus beginning to broaden at about the same distance from the umbo. Shell ornament consists of approximately 15 strong, fairly well incised angular costae on each valve, four of which occupy the ventral sulcus and extend to meet four or five on the dorsal valve. In general outline the species tends towards a sub-quadrate morphology, being slightly broader than long and having a shallow trapezoidal linguiform extension. OWEN AND ROSE: JURASSIC BRACHIOPODS 505 Internal structure. The highly crystalline nature of the matrix has made examination of the internal structures impossible and the generic assignment of this species should be regarded as provisional. Remarks. The external morphology of this species is typical of the genus described by S. S. Buckman (1918, p. 43, pi. 13, fig. 7) as Gibbirhynchia , the type species of which, G. gibbosa , was emended by Ager ( 1954, p. 36). Although the type species occurs in the Middle Lias, the genus is known to range from the Lower Lias, jamesoni Zone, where it is represented by G. curviceps (Quenstedt, 1871) in Britain and Germany. The species described here as Gibbirhynchia correcta (Di Stefano) bears a strong resemblance to G. curviceps in both size and in general outline. It differs from Quenstedl’s species, however, in its less acute sub-quadrate general outline, slightly produced umbo, longer or more extensive linguiform extension, and has fewer costae and a less well developed dorsal fold. The same distinguishing features separate it from Middle Liassic species, such as G. northamptonensis (Davidson, 1878), G. thorncombiensis (Buckman, 1922) and G. tiltonensis Ager, 1962 which appear to be confined to northern European facies (Ager 1962). Genus pontaltorhynchia gen. nov. Derivation of name. From Pontalto, central Apennines, Italy; NHM specimens from this locality are better preserved than those from Sicily or Gibraltar. Type species. Rhynchonella schopeni Di Stefano, 1886, p. 68, pi. 2, figs 45-46, from the Sinemurian of Sicily. Diagnosis. Transversely oval, multicostate, equibiconvex rhynchonellid. Umbo massive, beak short, sub-erect. Beak ridges rounded, indistinct, bordering a short interarea. Deltidial plates not exposed. Dorsal fold poorly developed, ventral sulcus shallow, trapezoidal linguiform extension moderate. Remarks. Bearing a superficial resemblance to species of Tetrarhynchia , Quadratirhynchia and Pseudogibbirhynchia , it can be distinguished from these genera in its shorter beak, rounded beak ridges and less angular or more rounded costae which are not so deeply incised. The internal characters as seen in the transverse serial sections are also distinct; described here for the type species. Pontaltorhynchia schopeni (Di Stefano, 1886) Plate 1, figures 12-17; Text-figures 2g-h, 3 1886 Rhynchonella schopeni Di Stefano, p. 68, pi. 2, figs 45a-c, 46a-c. 1952 Rhynchonella schopeni Di Stefano; Muir-Wood, in Bailey, p. 164. Material. NF1M specimens B85419, 85426-85427, 85439 from Gibraltar; B14969-14970 from Pontalto, Italy; and BB20288-20290 from Cape Taormina, Sicily; plus a single ONCP specimen, EM20313, from Gibraltar. Diagnosis. As for genus. Description. Medium-sized rhynchonellid, transversely oval in general outline. Lateral profile elongate-oval, anterior commissure broadly uniplicate with shallow linguiform extension. Mean dimensions approximately 1 9- 1 mm long, 2 1 -0 mm wide and 1 4-9 mm deep. Shell ornament of 25-28 strong, sub-angular to rounded costae originating from the umbo on each valve, with six to eight in the ventral sulcus and a corresponding number on the almost imperceptible dorsal fold. Short triangular planareas bounded by poorly defined beak ridges form the low apex of the shell. Internal structure. Description of the internal structures of this species is based on transverse serial sections through the umbo of a specimen collected from the Lower Lias, jamesoni Zone on the foreshore east of Cape Taormina, Sicily, by Dr M. K. Howarth (Text-fig. 3). It has been compared and agrees favourably with the figure given by Di Stefano (1886, pi. 2, figs 45-46) for Rhynchonella schopeni from the Calcare grigio, Gullo near St Antonio, Cape Taormma. The current location of this type material is unknown. 506 PALAEONTOLOGY, VOLUME 40 text-fig. 4. Twenty-four transverse serial sections through the umbo of a specimen of Merophricus mediterranea (Canavari, 1884) showing the pedicle collar, flat cardinal process, short ventrally concave hinge plates and low arcuate transverse band of the brachial loop; NHM BB 13873; Lower Lias of Ait-Oufella, High Atlas Mountains, Morocco. Scale bar represents 5 mm. No pedicle collar is developed within the ventral umbo. Sub-parallel dental plates develop early and support deeply inserted sub-quadrate hinge teeth. A strong high dorsal septum supports a deep septalium which diminishes rapidly. Moderately long hinge plates are ventrally deflected, their ventral surfaces becoming slightly convex. The hinge plates get thinner anteriorly developing into elongate crural bases. It was not possible to observe further internal structures beyond the tenth section due to the crystalline nature of the mhlling matrix. Remarks. Ager (1954, p. 30), in a description of the genus Gibbirhynchia , drew attention to what he considered to be the close relationship between that genus and the genera Tetrarhynchia, Quadratirhynchia and Grandirhynchia which range within the British Lias. Although external morphological aspects suggest a possible relationship, the internal characters of the specimen sectioned and described here as Pont alt orhynchia schopeni (Di Stefano) have little in common with other Liassic species. Serial sections of the type species of Tetrarhynchia figured by Ager (1956, p. 7, fig. 7) show an absence of a pedicle collar, a short septalium, moderately short hinge plates and elongate crural bases and thus could have been derived from the same original stock. Quadratirhynchia and Grandirhynchia share none of these internal characters with Pontaltorhynchia. OWEN AND ROSE: JURASSIC BRACHIOPODS 507 Superfamily terebratulacea Gray, 1 840 Family terebratulidae Gray, 1840 Subfamily plectoconchinae Dagis, 1974 Genus merophricus Cooper, 1983 Type species. Merophricus dubarii Cooper, 1983, p. 113, pi. 54, figs 14-15 ( = Terebratula cf. semiarata Dubar, 1942, p. 63, pi. 3, fig. 26a-e), from the Lias of the High Atlas Mountains, Morocco. Merophricus mediterranea (Canavari, 1884) Plate 1, figures 1-1 1 ; Text-figure 4 1880 Terebratula fimbrioides E. Deslongchamps; Canavari, p. 13, pi. 2, figs 1-2. 1881 Terebratula fimbrioides E. Deslongchamps; Canavari, p. 182, pi. 9, fig. lOa-d. 1884 Terebratula mediterranea Canavari, p. 85. 1935 Terebratula mediterranea Canavari; Dubar and Le Maitre, p. 9. 1942 Terebratula mediterranea Canavari; Dubar, p. 56, pi. 3, fig. 12a-d. 1952 Terebratula mediterranea Canavari; Muir-Wood, in Bailey, p. 164. 1952 Terebratula mediterranea var. elongata Dubar; Muir-Wood, in Bailey, p. 164. 1952 Terebratula mediterranea var. pectita Dubar; Muir-Wood, in Bailey, p. 164. Material. NHM specimens B85420, 85422, 85430-85438, 85440-85441 (Smith/Silvertop collection), B10409, 19524 (Goldie/Evans collection), and BB1 1504— 1 1505 (Alexander collection) from Gibraltar; and BB 13868-1 3875 from the Upper Sinemurian of Ait-Oufella, Morocco. Diagnosis. Medium to large, biconvex, polyplicate terebratulid with suberect beak and distinct beak ridges. Description. Shell elongate-oval in general outline, acutely biconvex with the dorsal valve slightly deeper than the ventral valve. Triangular conjunct deltidium obscured by a slightly incurved beak which is truncated by a large circular foramen. Lateral commissures straight or rectimarginate: anterior commissure varies from rectimarginate to slightly uniplicate. Broad anterior uniplica tion of the shell margins subject to polyplications which vary from coarse to moderately fine pseudocostation, originating from a point just anterior to the midvalve area. Internal structures. We have been unable to gain access to type material from the Lower Lias of the central Apennines: none could be located in the Canavari collection at the Department of Earth Sciences of the University of Pisa, Italy (L. Ragaini, pers. comm. 1996). Comparison has therefore been made between NHM specimen BB 13873, taken from a series collected from Morocco by the late R. V. Melville, the illustrations of Canavari (1881, pi. 9, fig. lOa-d), and also those of Dubar (1942, pi. 3, fig. 1 2a— d ). A series of transverse serial sections made through the umbo of the Moroccan specimen is illustrated here as Text-figure 4. The series shows a pedicle collar present, within the first 2 mm of the umbo. The conjunct deltidial plates are visible in transverse section and a flat, poorly developed cardinal process is evident in the dorsal umbo. Short triangular, concave, dorsally deflected hinge plates give rise to transversally sub-quadrate descending branches of the brachial loop which quickly develop long, inwardly curving processes which diminish anteriorly and finish in a broad but low arcuate traverse band. Remarks. Canavari (1880, p. 13) described a terebratulid from the Lower Lias of the central Apennines, assigning it to a species described by Eudes Deslongchamps (1855, 1862-85) as Terebratula fimbrioides from the Jurassic of Sarthe, Normandy. The specimens figured by Canavari (1880, pi. 2, figs 1-2) under the name T. fimbrioides, whilst resembling the specimens figured by Deslongchamps in having marginal plicae more strongly developed on the anterior part of the shell, do not agree in general outline, convexity or umbonal features. The specimens figured by Deslongchamps tend towards less acute biconvexity and a somewhat more pentangulate general outline. 508 PALAEONTOLOGY, VOLUME 40 Canavari (1881, p. 6, pi. 9) again described and figured a specimen as Terebratula finibrioides Deslongchamps, but this time he figured a different specimen (fig. 10) from the Lower Lias of the central Apennines which was more acutely biconvex, with a faint uniplication of the anterior margin and with strong or well developed marginal plicae. In a further description of brachiopod species from the same area Canavari (1884, p. 85) described, without illustration, a specimen under the new taxon Terebratula mediterranea and assigned his two specimens previously described under the name Terebratula fimbrioides E. Deslongchamps to the synonymy. Selection of a lectotype for the species is deferred whilst search for the specimens currently missing from the Canavari brachiopod collection at Pisa continues. Dubar (1942), in a description of Lower Liassic polyplicate Zeilleriidae and Terebratulidae from the Atlas Mountains and nearby regions, described and figured a specimen of a plicate terebratulid as Terebratula mediterranea Canavari (Dubar 1942, p. 42, pi. 3, fig. 12a-d). This specimen compares favourably with the specimen figured by Canavari (1881, pi. 9, fig. lOa-d) and also has many features recognized in Gibraltar specimens of polyplicate terebratulids. Among the many varieties of Terebratula mediterranea erected by Dubar (1942) is one which he described as Terebratula mediterranea var. pectita from Ait-Oufella, near Itzer (pi. 4, fig. la-c). A NHM specimen (B85440) comparable to this variant, collected from Gibraltar, is figured here (PI. 1, figs 1-3). It is a large specimen, 29 5 mm long, 20 5 mm wide and 19-7 mm deep. Dimensions of other Gibraltar specimens are: B85434, 23-0 mm long, 20-2 mm wide, 17-8 mm deep; B85435, 20-3 mm long, 18-8 mm wide, 14T mm deep; B85437, 19-8 mm long, 18-7 mm wide, 171 mm deep; B85438, 18 0 mm long, 16-2 mm wide and 14-2 mm deep. The species is here assigned to the genus Merophricus, hitherto known only from the type and the two closely similar originally assigned species M. semiarata (Dubar, 1942) and M. moreti (Dubar, 1942), on the basis of external morphology together with characteristic cardinalia and broad transverse band of the brachial loop. The brachial loop differs from those known in all other post- Palaeozoic terebratulacean genera (Cooper 1983). Superfamily zeilleriacea Allan, 1940 Family zeilleriidae Allan, 1940 Genus calpella gen. nov. Derivation of name. From Calpe, the classical name for Gibraltar. Type species. Zeilleria aretusa Di Stefano, 1886, p. 93, pi. 2, fig. 54, from the Lower Lias of St. Antonio, Taormina, Sicily. Diagnosis. Small polyplicate, sub-pentagonal zeilleriid. Remarks. Since its original description by Di Stefano, authors have consistently accepted the type species of Calpella as a zeilleriid. However, the genus can be distinguished from other Zeilleriidae by its shell ornament of strong plicae which, in some species, tend to be more highly developed marginally. Unlike Zeilleria sensu stricto it has a short interarea which is slightly concave and has marked beak ridges which are not persistent, nor are there any lateral depressions, as in Antiptychina , or anterior sulcation of either valve. Calpella aretusa (Di Stefano, 1886) Plate 1, figures 21-23 1886 Zeilleria aretusa Di Stefano, p. 93. pi. 2, fig. 54a-d. 1935 Zeilleria arethusa [.v/t] Di Stefano; Dubar and Le Maitre, p. 9. OWEN AND ROSE: JURASSIC BRACHIOPODS 509 1942 Zeilleria cf. tenuiplicata Dubar, p. 45, pi. 1, figs 23a-c, 24a-d. 1952 Zeilleria aff. tenuiplicata Dubar; Muir-Wood, in Bailey, p. 164. Material. NHM specimens B85420-85421 from Gibraltar. Diagnosis. As for genus. Description. The best preserved of these small zeilleriids is 15 mm long, 13-3 mm wide and 9-2 mm deep, sub- triangular in general outline, polyplicate with 10-12 well-developed plicae originating from the midvalve area and radiating anteriorly. Lateral margins are straight and the anterior margin rectimarginate, profile oval. The umbo is short and the beak suberect. Permesothyrid beak ridges border a short, slightly concave triangular interarea. The foramen is comparatively large. Internal structures. These have not been determined by serial sections because of the crystalline nature of the matrix. They are thought to be typically zeilleriid in form since a marked median septum can be seen through the dorsal valve extending from the umbo to a point just over half the length of the shell. Remarks. In general outline, plication and size, the specimen described and figured here (PI. 1, figs 21-23) has very much in common with the specimen figured by Di Stefano (1886, pi. 2, fig. 54a-d). It differs from that specimen, however, in its less acute biconvexity. It can be compared to a specimen figured by Dubar (1942, pi. 1, fig. 24a-d) from the Lower Lias of Tisdadine near Timhadit, Morocco which he included in the description of a series of polyplicate Zeilleriidae. This is very close to the specimen figured by Di Stefano (1886) as Zeilleria aretusa and shows the same degree of biconvexity of the valves and a very similar general outline. A further specimen figured by Di Stefano (1886) as Zeilleria aretusa shows the same degree of biconvexity of the valves and a very similar general outline. A specimen figured by Dubar as Zeilleria cf. tenuiplicata in the same publication (Dubar 1942), although more elongate and acutely biconvex than the specimen figured here (PI. 1, figs 21-23) nevertheless bears a strong resemblance to that specimen and is therefore included in the synonymy. It is probable that another specimen figured by Dubar (1942, pi. 1, fig. 23a-b) also belongs to the species Calpella aretusa (Di Stefano) but this is represented by a larger, flatter and broader example. DISCUSSION The systematic account is based upon the 26 specimens discussed by H. M. Muir-Wood (in Bailey 1952) and eight additional specimens. All but four (B19525, 85417-85418, 85442) have been identified to species level, with revised generic assignments. Despite their revised nomenclature, the brachiopod taxa are consistent with an early Lias (Lotheringian = late Sinemurian) age, as recognized previously for Gibraltar brachiopods by Dubar and Le Maitre (1935), Dubar (1942), and Muir-Wood (in Bailey 1952). According to their labels, the Natural History Museum specimens are those obtained by Colonel Charles Silvertop and James Smith, identified and discussed although not formally described by Muir-Wood; three further specimens (one presented by J. W. Goldie prior to 1899, two by F. Evans in 1903) previously wrongly identified as Hemiptychina and of Carboniferous age (Reed 1949); and two specimens presented by Captain G. B. Alexander, effectively in 1948. Three specimens loaned by the Office National de Gestion des Collections Paleontologiques at the Universite Claude Bernard: Lyon 1 were those originally obtained by D. T. Ansted from the Austrian Consul-General on Gibraltar, sent to E. de Verneuil for identification, and subsequently donated to the Ecole des Mines in Paris. The exact provenance of the Silvertop/Smith specimens is not known : there is nothing other than ‘Gibraltar’ on the labels which accompany them and Smith (1846) in his reference to brachiopods gave no locality data. The matrix associated with the specimens is a crystalline carbonate, typical both of the Gibraltar Limestone Formation and of beds within the lower part of the Catalan Bay Shale Formation. No brachiopods have been collected from either formation in recent years. How- ever, Ansted (1859) and Hochstetter (1866) referred to brachiopods collected by Mr Frembly (the Austrian Consul-General) during four years of military quarrying on the north-east side of the 510 PALAEONTOLOGY, VOLUME 40 Rock, Ramsay and Geikie (1878) recorded that their ‘only good specimens’ (of brachiopods) all came from a quarry on the North Face, and Dubar (in Dubar and Le Maitre 1935) also seems to have obtained brachiopods from the North Face. Since the Catalan Bay Shales were not exposed either at the North Face or in Catalan Bay along the east coast until much later and after more extensive military quarrying (Rose and Rosenbaum 1991 b, pp. 79, 167), it is almost certain that it was the Gibraltar Limestone Formation which yielded all these brachiopods, and that in consequence at least part of it can be dated as Sinemurian. The Goldie/Evans specimens are more precisely sourced, to Catalan Bay Quarry. Although this quarry is now the type locality for the Catalan Bay Shales, the Shales were not exposed until the quarry was extended in 1943 to provide fill for the expansion of the airfield. At the time the brachiopods were collected (prior to 1903), only the Gibraltar Limestone and Quaternary screes derived from it in the cliffs above were accessible. The near vertical clifi's tower to a height of some 300 m above the quarry floor, exposing a faulted succession in which all four members of the Gibraltar Limestone Formation are represented. The horizon yielding the brachiopods has not yet been identified. The Alexander specimens are recorded as from the ‘searchlight scree, above Arow Street, 300 feet above the base of the limestone’. The dolomitic lower members of the Gibraltar Limestone Formation reveal few fossils other than oncolitic or stromatolitic algae. However, molluscs are increasingly abundant higher in the succession. Disarticulated bivalves are locally common at horizons within both the Keightley and Buffadero members (Rose and Rosenbaum 1991, figs 6.10, 6.13c), and gastropods (Rose and Rosenbaum 1991, fig. 6.13) are locally common within the Buffadero Member. Coral and echinodern fragments are also known from this horizon. Unfortunately, the rock is very well cemented and the fossils are seen only in cross section, so cannot be identified precisely, or easily extracted. A large gastropod variously identified as Eulima hedingtonensis Sowerby or as Pseudomelania (by Choffat 1892), originally in the James Smith collection at the Geological Society of London, seems not to have survived the transfer of most of the Society’s collection to The Natural History Museum, London (Dr N. J. Morris, pers. comm. 1994). Dubar and Le Maitre (1935) referred to beds of Chemnitzia at the North Face, possibly the same taxon. The Gibraltar Museum contains the internal mould of a large pleurotomariid gastropod, labelled as from Little Bay, but its matrix is not that of the Gibraltar Limestone. However, since the specimen cannot be more precisely identified and therefore dated (N. J. Morris, pers. comm.), it is of little significance. Smaller, thick-shelled cerithiid gastropods are commonly seen in cross section in the higher beds of the Limestone, where their increasing abundance coincides with a decrease in abundance of the algal laminites and presumably a change to less restricted marine environments. Stromatomorpha liasica Le Maitre, described by Dubar and Le Maitre (1935) as a spongiomorphid but conforming to the currently accepted definition (Wood 1987) of stromatoporoid sponges, is known only from the type specimens, collected from the North Face. A fossil, ‘apparently a nautilus’, was observed in a building stone by Duckworth (1911, p. 355), but the identification was tentative, and the source of the stone uncertain. The building has now been demolished and the specimen lost. Thus as yet the brachiopods provide the only means of dating. That the Gibraltar Limestone Formation of the main ridge area of the Rock has been overturned can be demonstrated convincingly both from algal growth directions and from statistical analysis of geopetal infills of the numerous gastropod shells. The Catalan Bay Shale Formation which crops out beneath the overturned Gibraltar Limestone Formation at the base of the North Face and along parts of the east coast of the Rock should therefore have a younger date. Although the ammonites to which L. F. Spath (in Bailey 1952) ascribed a Mid Lias (Domerian, late Pliensbachian) age have now been lost, M. K. Howarth (in Rose and Rosenbaum 1990) has confirmed that in his identifications and discussion of these ammonites, Spath was almost certainly correct in his age assignment. Of the supposedly closest comparable carbonate outcrops, that at Los Pastores, 9 km westwards across the Bay of Gibraltar, is now largely quarried away, tectonically complex, and of broader but imprecise Triassic-Jurassic date (Valenzuela Tello 1993). A thick, well-exposed sequence of shallow- OWEN AND ROSE: JURASSIC BRACHIOPODS 511 water carbonates associated with cherty 'shales’ occurs immediately north-east of Manilva, Spain, some 30 km to the north of Gibraltar, and has been mapped as of Liassic age (Fontbote 1970). Approximately 24 km southwards across the Strait of Gibraltar, Gebel Musa (which in classical times was twinned with Gibraltar under the name 'Pillars of Hercules’) has also been described as a klippe dominantly of Early Jurassic carbonates (Durand-Delga and Villiaumey 1963). However, so far as we are aware, none of these localities has yielded a described brachiopod fauna. The closest faunal comparisons currently possible are with the Liassic faunas of the High Atlas in Morocco (Dubar 1932, 1942), and of Taormina in eastern Sicily (Di Stefano 1886). Of the five species described here from Gibraltar, three (Gibbirhynchia correcta , Pont alt orhynchia schopeni , Calpella aretusa) were first described from Sicily (Di Stefano 1886), one ( Merophricus mediterranea) from the central Apennines (Canavari 1884), and the fifth (Liospiriferina rostrata ) as a Sicilian species here merged in synonymy. The new genus Pontaltorhynchia is defined herein on the basis of material not only from Gibraltar but also Sicily and the central Apennines. Faunal affinities between Gibraltar and the Lower Lias of Italy are therefore very clear. Additionally, three of the species have been described from Morocco (as ‘ Terebratula' mediterranea and ‘ Zeilleria' aretusa , and Liospiriferina rostrata ), extending their apparent geographical range over 2000 km westwards. ' 77 mediterranea is here newly ascribed to the genus Merophricus , according to Cooper (1983, p. 1 14) a genus endemic to Morocco and with a distinctive morphology: a brachial ‘loop unlike any genus described’ in his massive, authoritative account of the post-Palaeozoic Terebratulacea (204 genera). ‘Z.’ aretusa is here made the type of the new genus Calpella. Both Merophricus and Calpella thus appear to be widespread in, but also endemic to, the Lower Lias of the western Mediterranean. Dubar and Le Maitre ( 1935, p. 9) correlated Gibraltar brachiopods (from low on the North Face) with those from the Sinemurian of Morocco, but failed to recognize that the sequence in the North Face was inverted. Consequently, their relative ages and correlations assigned to the beds bearing Chemnitzia and Stromatomorpha must be in error. An Early Jurassic age for much of the Gibraltar Limestone Formation is, however, consistent with that of the thick Bahamian-type platform carbonates widely developed along the southern continental margin of the Alpine-Mediterranean Tethys. Bernoulli and Jenkyns (1974) reviewed evidence for thick sequences of Liassic shallow-water carbonates in the Subbetic region of Spain, in the northern Rif and High Atlas regions of Morocco, in Tunisia, and in Sicily and several regions of mainland Italy. As in Gibraltar these sequences terminate with a sharp change to more pelagic sedimentation in the mid-late Lias, a change that Bernoulli and Jenkyns correlated with widespread block faulting, probably related to the onset of rifting in the oceanic Tethys. Acknowledgements. We are grateful to A. Prieur for the loan of three specimens by the Office National de Gestion des Collections Paleontologiques at the Universite Claude Bernard: Lyon 1, Villeurbanne, France; to Dr J. C. Finlayson for loan of three gastropods (two now identified as Quaternary) from the Gibraltar Museum; to Drs N. J. Morris and R. Wood for comment on the affinities of Gibraltar gastropods and spongiomorphids respectively; to Dr Luca Ragaini for attempts to locate Canavari type specimens at the University of Pisa; and to Prof. Michael Sandy for helpfully reviewing the manuscript. REFERENCES ager, d. v. 1954. The genus Gibbirhynchia in the British Domerian. Proceedings of the Geologists' Association , 65, 25-51. — 1956. The British Liassic Rhynchonellidae. Part 1. Monograph of the Palaeontographical Society, 110 (476), i-xxvi, 1-50, pis 1-4. — 1962. The British Liassic Rhynchonellidae. Part 3. Monograph of the Palaeontographical Society , 1 14 (498), 85-136, pis 8-11. — 1965. Mesozoic and Cenozoic Rhynchonellacea. H611. In moore, r. c. (ed. ). Treatise on invertebrate paleontology. Part H. Brachiopoda. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 927 pp. 512 PALAEONTOLOGY, VOLUME 40 allan, r. s. 1940. A revision of the classification of the terebratelloid Brachiopoda. Records of the Canterbury Museum , 4, 267-275. ansted, d. t. 1859. On the geology of Malaga and the southern part of Andalusia. Quarterly Journal of the Geological Society , London , 15, 585-604. arkell, w. J. 1956. Jurassic geology of the world. Oliver & Boyd, Edinburgh, xv + 806 pp. bailey, E. b. 1952. Notes on Gibraltar and the northern Rif. Quarterly Journal of the Geological Society , London , 108. 157-175. Bernoulli, d. and jenkyns, h. c. 1974. Alpine, Mediterranean, and central Atlantic Mesozoic facies in relation to the early evolution of the Tethys. In dott, r. h. and shaver, r. h. (eds). Modern and ancient geosynclinal sedimentation. Special Publication of the Society of Economic Paleontologists and Mineralogists , 19, 129-160. buckman, s. s. 1918. The Brachiopoda of the Namyau Beds, northern Shan States, Burma. Memoirs of the Geological Survey of India. Palaeontologia Indica, New Series 5, Memoir 2, 1-299, pis 1-21. [Imprinted 1917], 1922. Jurassic chronology II. Certain Jurassic strata near Eypesmouth (Dorset). Quarterly Journal of the Geological Society , London , 78, 378-457. canavari, m. 1880. I Brachiopodi degli strati a Terebratula Aspasia Mgh. nelf Appennino Centrale. Atti della Reale Accademia dei Lincei. Memorie della classe di scienze fisiche , matematiche e naturali, Serie 3 , 8, 1-34, pis 1-4. 1881. Alcuni nuovi Brachiopodi degli strati a Terebratula Aspasia Mgh. nelf Appennino Centrale. Atti della Societa Toscana di Scienze Naturali , 5(1), 1-12, pi. 9. 1884. Contibuzione III alia conoscenza dei Brachiopodi degli strati a Terebratula Aspasia Mgh. nelf Appennino Centrale. Atti della Societa Toscana di Scienze Naturali , 6 (9), 72-107, pis 9—11. choffat, p. 1892. Sur Page du Rocher de Gibraltar. Bulletin de la Societe Geologique de France , Serie 3 , 20. ix-x. cooper, g. a. 1983. The Terebratulacea (Brachiopoda), Triassic to Recent: a study of the brachidia (loops). Smithsonian Contributions to Paleobiology, 50, 1-445. dagis, a. s. 1974. Triasovie Brakhiopody (Morfologia, Sistema, Filogenii, Stratigraficheskoe Znachenie i Biogeografia). Trudy, Institut Geologii i Geofiziki, Siberskoe Otdelenie, Akademia Nauk SSSR, 214, 1-386, pis 1-49. [In Russian]. davidson, t. 1878. The British fossil Brachiopoda. Vol. 4, Part 2, No. 2. Monograph of the Palaeontographica! Society, 32 (146), 145-242, pis 17-29. 1884. The British fossil Brachiopoda. Vol. 5. Part 3. Monograph of the Palaeontograpliical Society, 38 (183), 243-476, pis 18-21. deslongchamps, E. e. 1855. Note sur deux nouvelle terebratules du Lias moyen de Precigne (Sarthe). Memoires de la Societe Linneenne de Normandie, 10, 1-4, pi. 17. 1862-85. Paleontologie frangaise on description des animaux invertebres fossiles de la France: Terrain jurassique 1 . Brachiopodes. Masson et Fils, Paris, 448 pp., 131 pis. di stefano, G. 1886. Sul Lias inferiore di Taormina e de’ suoi dintorni. Giornale della Societa di Scienze Naturale ed Economiche di Palermo, 18, 46-184. dubar, G. 1932. Les Lias et le Jurassique de la Haute Moulouya et du Haut Atlas (sud et sud-est de Midelt). Bulletin de la Societe Geologique de France , Serie 5, 2, 573-594. 1942. Etudes paleontologiques sur le Lias du Maroc. Brachiopodes, Terebratules et Zeilleries multiplissees. Notes et Memoires du Service des Mines et de la Carte Geologique du Maroc, 57, 1-103. and le maitre, d. 1935. Etudes paleontologiques sur le Lias du Maroc. Spongiomorphides et Algues. Notes et Memoires du Service des Mines et de la Carte Geologique du Maroc , 34, 1-58. duckworth, w. l. H. 1911. Cave exploration at Gibraltar in September, 1910. Journal of the Royal Anthropological Institute , 41, 350-380, pis 40-43. dumeril, A. M. c. 1806. Zoologie analytique, ou methode naturelle de classification des animaux. Allais, Paris, xxiv + 344 pp. durand-delga, m. and villiaumey, m. 1963. Sur la stratigraphie et la tectonique du groupe de Gebel Musa. Bulletin de la Societe Geologique de France, Serie 7, 5, 70-79. fontbote, j. m. 1970. Mapa geologica de Espaha. Scale l : 200,000. Sheet 87 Algeciras. Instituto Geologico y Minero de Espafia, Madrid. gemmellaro, G. G. 1874. Sopra alcune faune giuresi e liasiche della Sicilia. Studi paleontologici. 3: Sopra i fossili della zona con Terebratula Aspasia Menegh. della provincia di Palermo e di Trapani. Giornale della Societa di Scienze Naturali ed Economiche di Palermo 10, 53-112, 3 pis. OWEN AND ROSE: JURASSIC BRACHIOPODS 513 gray, J. E. 1840. Synopsis of the contents of the British Museum. 42nd edition. British Museum, London, 370 pp. 1848. On the arrangement of the Brachiopoda. Annals and Magazine of Natural History, Series 2, 2, 435 — 440. hochstetter, F. von 1866. Geologische Beobachtungen wahrend der Reise der Osterreichischen Fregatte Novara um die Erde. Geologischer Theil, 2 (1). Aus der Kaiserlich-Koniglichen Hof- und Staatsdruckerei, Wien, xiv+ 161 pp. huxley, t. h. 1869. An introduction to the classification of animals. John Churchill and Sons, London, 147 pp. quenstedt, F. A. 1868-71. Petrefactenkunde Deutschlands. 2, Brachiopoden. Fues’s Verlag, Leipzig, 748 pp. atlas pis 37-61 . ramsay, A. c. and geikie, j. 1878. On the geology of Gibraltar. Quarterly Journal of the Geological Society, London. 34. 505 — 4 1 . reed, f. r. c. 1949. The geology of the British Empire. 2nd edition. Arnold, London, ix + 764 pp. roemer, f. 1864. Geologische Reise Notizen Spanien. Neues Jahrbuch fur Mineralogie, Geologic und Paldontologie, 1864, 769-806. rose, e. p. f. and rosenbaum, m. s. 1990. Royal Engineer geologists and the geology of Gibraltar. Part III. Recent research on the limestone and shale bedrock. Royal Engineers Journal, 104, 61-76. 1991«. The Rock of Gibraltar. Geology Today, 7, 95-101. — 19916. A field guide to the geology of Gibraltar. The Gibraltar Museum, Gibraltar, 192 pp. — 1994a. The Rock of Gibraltar and its Neogene tectonics. Paleontologia i Evolucid, 24-25, 41 1-421. — 19946. The pre-Quarternary geological evolution of Gibraltar. In rodriguez vidal, j., diaz del olmo, f., finlayson, c. and Giles pacheco, F. (eds). Gibraltar during the Quaternary. AEQUA Monografias, Seville, 2, 6-1 1 . rosenbaum, m. s. and rose, e. p. f. 1991. Geology of Gibraltar. Single sheet 870 x 615 mm. Side 1 cross-sections and bedrock geology map 1 : 10,000, Quarternary geology, geomorphology and engineering use of geological features maps 1 : 20,000; Side 2 illustrated geology (combined bedrock/Qua ternary geology) map 1 : 10,000, plus 17 coloured photographs/figures and explanatory text. School of Military Survey Miscellaneous Map 45. rouselle, L. 1977. Spiriferines du Lias moyen et superieur au Maroc (Rides Prerifaines; Moyen Atlas) et en Espagne (Chaine Celtiberique Orientale). Notes et Memoires du Service Geologique du Maroc, 268, 153-174, pi. I. SCHLOTHEIM, E. F. von 1820-23. Die Petrefactenkunde auf ihrem jetzigen Standpunkte durcli die Beschreibung einer Sammlung versteinerter und fossiler Uberreste der Tier- und Pflanzenreichs der vorvelt erldutert. Becker, Gotha, 726 pp., atlas 37 pis. smith, j. 1846. On the geology of Gibraltar. Quarterly Journal of the Geological Society, London, 2, 41-51. Valenzuela tello, J. M. 1993. El aflorcuniento paleontologica de “ Los Pastores ”, Algeciras (Cadiz). Instituto de Estudios Campogibralterenos, Algeciras, 145 pp. wood, r. 1987. Biology and revised systematics of some late Mesozoic stromatoporoids. Special Papers in Palaeontology , 37, 1-89. zieten, C. H. 1830-33. Die Versteinerungen Wurtembergs. Verlag der Expedition des Werkes unserer Zeit. Stuttgart, 102 pp., 72 pis. ELLIS F. OWEN Department of Palaeontology The Natural History Museum London SW7 5BD, UK EDWARD P. F. ROSE Department of Geology Typescript received 14 June 1996 Royal Holloway, University of London Revised typescript received 15 October 1996 Egham, Surrey TW20 0EX, UK LIFE HISTORIES OF SOME MESOZOIC ENCRUSTING CYCLOSTOME BRYOZOANS by frank k. McKinney and Paul d. taylor Abstract. Single-layered, multiserial cyclostome bryozoans are almost ubiqitous as encrusters of Mesozoic hard substrata but little attention has been paid previously to the attributes of their life histories obtainable from their fossil skeletons. Colonies from ‘populations’ of one Triassic, five Jurassic and nine Cretaceous species from England and Slovakia are here studied using an image analyser to record colony size and shape, and the number, location and sizes of larval brood chambers. Survivorship curves relative to colony size demonstrate varying patterns of mortality for different species. None of the species shows evidence of a fixed maximum colony size. Some species were capable of producing frontal, or more commonly, peripheral subcolonies. These species typically have smaller colonies than species without subcolonies. Colony size at the onset of female sexual reproduction was found to be relatively constant in some species but variable in most, possibly indicating that an environmental cue triggered reproduction. Most colonies reproduced only once (semelparity) and apparently died shortly afterwards, but a few survived to reproduce a second time (iteroparity). No correlation among species was found between skeletal measures of reproductive effort and colony size. Flexibility in life history patterns predominate in the 15 studied species, the one notable exception being Actinopora disticha which was relatively deterministic. Senescence is not a necessary attribute of the life history of clonal organisms (Sackville Hamilton et al. 1987 and references therein; Orive 1995), and indeed the ages attained by some plants, corals, and bryozoans have been shown to be very great, up to at least 13000 years for plants (Cook 1983). These organisms show no signs of clonal senescence and are, at least in comparison with solitary organisms, potentially immortal except for the ever-present possibility of death from extrinsic, environmental causes. Despite apparent potential immortality, clonal organisms exhibit a wide spectrum of life histories. Some have strongly deterministic growth and may have brief, truncated life-spans, sometimes with clonal senescence (e.g. Chadwick-Furman and Weissman 1995), while others may be longer-lived and have chaotic growth patterns that are disrupted locally by small-scale biotic and abiotic environmental perturbations. Some clonal organisms track an environment through directed growth ; for example, they are capable of maintaining local growth trajectories indefinitely within long-lived stationary or slowly migrating environments (e.g. Lasker 1983). However, most clonal taxa fall between such extremes and have their life-spans determined by seasonal changes or limited by site-specific interactions or disturbances. Bryozoans are exclusively clonal, and they are common to abundant in many marine benthic environments. Among Recent calcified bryozoans, the gymnolaemate order Cheilostomata is by far the more diverse, abundant, and conspicuous. Consequently, cheilostomes have been studied more frequently relative to other bryozoans and much more is known about aspects of their life histories (Gordon 1970; Eggleston 1972; Dudley 1973; Hayward and Ryland 1975; Mawatari 1975; Dyrynda and Ryland 1982; Winston and Jackson 1984; Jackson and Wertheimer 1985 ; Harvell and Grosberg 1988; Karande and Udhayakumar 1992; Hunter and Hughes 1993; Bishop 1994). The other calcified bryozoans, the stenolaemate order Cyclostomata, are less diverse, abundant, and conspicuous than cheilostomes in Recent communities. Life-history information on living cyclostome bryozoans is limited and consists predominantly of data on embryo formation and larval longevity for a few species (Harmer 1896; Borg 1926). Data on growth rates (Vail and Wass (Palaeontology, Vol. 40, Part 2, 1997, pp. 515-556, 5 pis) © The Palaeontological Association 516 PALAEONTOLOGY, VOLUME 40 198 \b), longevities (Winston 1985), relationships between colony growth habits and reproduction (Taylor 1979; McKinney 1983), and reproductive schedules (Ryland 1963) are extremely sparse for living and fossil stenolaemates. Life histories of extinct benthic marine organisms represented in the fossil record cannot be known to the same level of detail as for living organisms. With very few exceptions, such as daily to annual banding in corals (e.g. Johnson and Nudds 1975), there are few practical ways to measure growth rates of individuals. Furthermore, population parameters are difficult to obtain because it is seldom possible to be sure that fossil skeletons preserved on a single bedding plane were actually coeval (Flessa et al. 1993) and there are several processes that, potentially, can selectively remove skeletons, resulting in biased size distributions (see below). However, given the limitations imposed by working with death assemblages that may span an unknown amount of time, one can determine certain aspects of life history if taphonomic effects have not biased the sample. Where size can be taken as an adequate surrogate for age, survivorship curves can be constructed. Colony shape, and capacity and frequency of asexual reproduction can be determined, and, where sexual reproduction is reflected in skeletal structures, size at reproduction, aspects of reproductive effort, and survival beyond reproduction can be quantified. A variety of growth habits has been produced by bryozoans, which reflect functional, phylogenetic, and life-history differences (McKinney and Jackson 1989). Single-layered, encrusting, multiserial cyclostome colonies belonging to diverse species are common on small hard substrata, such as skeletons of other benthic invertebrates, lithic clasts and hardgrounds, particularly in the Mesozoic (e.g. Brood 1972; Holder 1972; Palmer and Fiirsich 1974, 1981 ; Mayoral and Sequeiros 1981 ; Wilson 1986; Walter 1989; Palmer and Wilson 1990; Martill and Hudson 1991 ; Taylor and Michalik 1991; Bertling 1994). These colonies generally grow radially by budding new zooids around their perimeters. Spread of the colony is commonly interrupted by microenvironmental conditions including crowding and competitive contacts, predation or small-scale obstructions. Colonies in some species develop subcolonies usually along their periphery. Colonies that have not generated subcolonies are here referred to as ‘solitary’ colonies. Those that have generated subcolonies are referred to as ‘compound’ colonies, with the original portion referred to as the ‘parent’ colony and each subcolony as a ‘daughter’. Subcolonies originate from a small number of zooids in the parent colony and may either grow away from the parent colony or develop radial growth and overlap the parent colony. In a few species, subcolonies may develop from local regions within the interior of the parent colony rather than from some point around the perimeter, so that as the subcolony develops it is ‘stacked’ above the parent. In this paper, life history-related characteristics of 15 species of encrusting, Mesozoic cyclostomes are described and compared, and their implications for the life-histories of these species are discussed. This is one of an informal series of papers that will build information on the diversity and range of cyclostome life histories through time so that eventually the pattern of long-term changes can be documented and analysed for constraints and possible causes. MATERIAL AND METHODS This study is based predominantly on suites of specimens from the collections of The Natural History Museum, London (BMNH); other figured or mentioned specimens are from the Sedgwick Museum, University of Cambridge (SM), and the Slovak National Museum, Bratislava, Slovakia (SNM). All available specimens of encrusting Mesozoic cyclostomes in the BMNH collections were examined, and suitably preserved species represented by a sufficient number of specimens and for which collecting bias was thought to be minimal were included in this study. Some of the specimens had been previously catalogued individually, but the majority were in uncurated samples derived from various sources. Most of the uncurated Cretaceous material was collected from the Chalk by C. T. A. Gaster and A. W. Rowe in the late 19th and early 20th centuries. Triassic and Jurassic specimens were collected by us. Several specimens of Plagioecia sp. 2 encrusting a single valve of mckinney and taylor: bryozoan life histories 517 table 1. Occurrence data for species included in this study. Species Location Stratigraphical horizon Number of specimens Actinopora disticha Northfleet, Kent coranguinum Zone 120 (von Hagenow) Liripora complanata (Roemer) Seaford and Offham Hill, Lewes, East Sussex cortestudinarium Zone 51 ‘ Mesonopora ' laguncula Northfleet, Kent coranguinum Zone 43 (Voigt) Plagioecia aff. carinata Northfleet, Kent coranguinum Zone 19 (Levinsen) Discocavea irregularis (d'Orbigny) Northfleet, Kent; Wanborough, Wiltshire; Eastbourne, and Seaford. East Sussex; West Horsley, Surrey; Norwich, Norfolk coranguinum Zone mucronata Zone 146 Hyporosopora dilatata (d’Orbigny) Stanton Harcourt, Oxfordshire Oxford Clay (Callovian/Oxfordian) 45 Warboys Clay Pit, Cambridgeshire Oxford and Ampthill clays (Oxfordian) 93 Plagioecia^. reniformis Northfleet, Kent coranguinum Zone 52 (Gregory) Eurystrotos aff. acanthina Northfleet, Kent coranguinum Zone 19 (Gregory) Hyporosopora sp. Stanton Harcourt, Oxfordshire Oxford Clay (Callovian/Oxfordian) 14 " Mesonopora ’ sp. Seaford and Offham Hill, Lewes, East Sussex; Luton, Chatham, Kent Northfleet, Kent cortestudinarium Zone coranguinum Zone 161 Plagioecia sp. 1 (zooids 200 //m width)* Stanton Harcourt, Oxfordshire Oxford Clay (Callovian/Oxfordian) 18 Plagioecia sp. 2 (zooids 100 pm width) Stanton Harcourt, Oxfordshire Oxford Clay (Callovian/Oxfordian) 14 Warboys Clay Pit, Cambridgeshire; St Ives, Cambridgeshire Oxford and Ampthill clays (Oxfordian) Oxford Clay 47 Plagioecia cf. disciformis (von Hagnow) Seaford and Offham Hill, Lewes, East Sussex cortestudinarium Zone 27 Reptomultisparsa hybensis (Prantl) Viper Pit, Hybe, Slovakia Hybe Beds (Rhaetian) 88 Reptomultisparsa sp. Stanton Harcourt, Oxfordshire Oxford Clay (Callovian/Oxfordian) 6 * Species illustrated in Martill and Hudson (1991, pi. 34, figs 3 and 5). Gryphaea dilatata (Sowerby) from the Oxford Clay of St Ives, Huntingdonshire (SM J26482) were also included in the study. Each of these collections of individual species was treated as a death assemblage that was unaffected or minimally affected by taphonomic processes or collecting bias. There may have been 518 PALAEONTOLOGY, VOLUME 40 table 2. Summary of colony size attributes. Measurements are in mm and mm2 and are given as mean/median. See Table 1 for number of specimens used for determining statistical characteristics. Diameters Curve relating area to max. diam. Species Max. Min. Ratio Area Actinopora disticha 3-6/3-5 29/2-9 0-81/0-83 9- 1/7-5 Y = 0-644X1 512 Liripora complanata 5-2/4-6 4-4/41 0-85/0-89 18-7/14-0 Y = 1-024X1 700 ' Mesonopora ' laguncula 120/10-8 9-8/10-0 0-82/0-93 102-8/77-6 Y = 1T60X1750 Plagioecia aff. carinata 13-2/ 13-3 9-6/10-1 0-73/0-76 99-6/88-6 Y = 0-383X2 111 Discocavea irregularis 3-5/2-8 3-1/2-5 0-89/0-89 10-5/5-5 Y = 0-715X1938 Hyporosopora dilatata 1 1-9/1 10 89/8-6 0-75/0-78 109-3/100-3 Y = 0-747X1 562 (Stanton Harcourt) Hyporosopora dilatata 8-2/ 5-5 6-2/44 0-76/0-80 494/21-9 Y = 0-699X1 851 (Warboys Clay Pit) Plagioecia ? reniformis 9-3 /8-9 7- 1/6-7 0-76/0-75 69-2/55-8 Y = 0-687X1 860 Eurystrotos aff. acanthina 4-3/4-3 3-6/3-7 0-84/0-86 13-0/12-5 Y = 0-751X1855 Hyporosopora sp. 8- 1/8-0 7-3 /7-3 0-90/0 91 47-1/33-5 Y = 0-685X1 975 ' Mesonopora' sp. 6-6/5-S 5-5 / 5-0 0-83/0-91 454/23-6 Y = 0-763X1 901 Plagioecia sp. 1 6-7/64 6-0/5-8 0-90/0-91 33-2/29-2 Y = 0-698X1 996 Plagioecia sp. 2 6-5/6- 1 48/4-7 0-74/0-77 23-9/21-7 Y = 1-100X1585 (Stanton Harcourt) Plagioecia sp. 2 4-0/3-6 34/3-1 0-85/0-86 12-2/8-9 Y = 0-67 5X2000 (Cambridgeshire) Plagioecia cf. disciformis 2-3/2-0 2-0/ 1 -7 0-87/0-85 44/24 Y = 0-649X2016 Reptomultisparsa hybensis 4-5/3-9 3-7/3-5 0-92/0-90 15-6/10 1 Y = 0-60 7X2005 Reptomultisparsa sp. 12-7/13-2 10-3/11-2 0-81/0-85 150 0/189-1 Y = 0-753X1 905 taphonomic loss of some of the smallest colonies through abrasive grazing by durophagous echinoids prior to burial, although grazing damage of the general shell surfaces on which the colonies grew was not conspicuous, and very small cyclostome colonies of one to a few zooids were found among the materials from most sites. Fine-grained matrix and whole fossils characterize the Cretaceous chalks, Jurassic clays, and Triassic silty marls from which the suites of specimens were collected, implying local origin of the fossils and low kinetic energy at the site of deposition. Taphonomic bias due to selective abrasion and transportation is therefore unlikely. The cyclostome colonies grew as permanent encrustations on solid calcitic substrata, including brachiopods, echinoids and ostreoid bivalves. Cyclostoines too have calcite skeletons (Poluzzi and Sartori 1975). In all lithologies from which suites of specimens were collected the calcite skeletons are well-preserved, so taphonomic bias due to selective dissolution of the smallest specimens is unlikely, especially inasmuch as the colonies were bonded to relatively large substrata of the same composition. In summary, although preferential loss of small colonies may have occurred so that the proportion of colonies in the smallest potential sizes is underrepresented (as is thought to be the case for most fossil assemblages; Craig and Oertal 1966), there is no evidence in any of the studied assemblages for selective removal of small colonies by durophagous grazers, nor from mechanical abrasion or transportation, nor preferential dissolution. The cyclostome assemblages are therefore treated as taphonomically unbiased, time-averaged death assemblages. Collecting bias is also thought to have relatively small impact on size distributions, with the following exception. A. W. Rowe and C. T. A. Gaster, who made the bulk of the Cretaceous collections, kept a wide variety of colony sizes and shapes, giving the appearance that they kept all mckinney and taylor: bryozoan life histories 519 table 3. Summary of colony reproductive attributes. Measurements are in mm and mm2 and are given as minimum/mean/median for ancestrula to ooeciopore and mean/median for ooeciopore to colony edge. Square brackets include number of observations. Species Brood chamber shape (width/length) Ancestrula to ooeciopore Ooeciopore to colony edge Values (SD) % area as gonozooids in fertile colonies %fertile colonies Actinopora disticha 7-1 0-6/11/1-0 0-4/0-4 (0-24) 5-4 34-6 [41] [41] [41] [27] [78] Liripora complanata 1-21 0-9/2-5/2-7 0-4/0- 3 (0-24) 1-1 219 [6] [7] [7] [7] [32] 1 Mesonopora ’ laguncula 0-9 2-8/4-5/4-4 15/1-4 (0-42) 2-8 32-2 [20] [20] [20] [10] [3lJ Plagioecia aff. car mat a 3-3 6-1/6 4/6-4 10/10 (-) 14 5-3 [2] [2] [2] [1] [19] Discocavea irregularis irregular 0-7/0-7/0-7 06/0-6 (-) 23-0 2-5 [2] [2] [2] [1] [40] Hyporosopora dilatata 10 2-6/6-3/6-7 1-8/14 (1-71) 2-6 26-1 (Stanton Harcourt) [29] [29] [29] [11] [42] Hyporosopora dilatata 1-1 2-2/6-0/4-9 1 8/1-8 (0-97) 3-3 23-3 (Warboys) [19] [19] [19] [10] [43] Plagioecia! reniformis 1-6 1-9/3-5/31 0-9/0-7 (0-78) 0-8 55 [59] [58] [58] [22] [40] Eurystrotos aff. acanthina 1-6 0-9/1-8/2-2 03/0-2 (-) 3-5 63-2 [19] [19] [19] [11] [16] Hyporosopora sp. 0-7 2-3/4-0/3-7 1 -4/1-2 (1-08) 6-0 57-1 [42] [42] [42] [8] [14] ‘ Mesonopora ' sp. 30 1-1/3-3/2-9 0-8/0-7 (0-62) 6-5 29-7 [94] [95] [97] [41] [138] Plagioecia sp. 1 30 2-1/3-1/2-9 M/0-9 (0-75) 6-8 88-2 [46] [46] [46] [16] [18] Plagioecia sp. 2 2-4 1-4/2-9/2-7 M/0-6 (1-28) 6-8 71-4 (Stanton Harcourt) [34] [34] [34] [10] [14] Plagioecia sp. 2 3 0 1-3/1 8/1-5 0-7/0-7 (0-24) 5-4 21 (Cambridgeshire) [9] [9] [9] [6] [34] Plagioecia cf. disciformis 2-5 10/1-3/1-3 0-3/0- 3 (0-05) 5-5 11-1 [4] [4] [4] [3] [277] Reptonmltisparsa hybensis 0-3 2-0/3-6/3-7 0-9/0-8 (0-54) 2-5 14-8 [12] [12] [12] [4] [27] Reptomultisparsa sp. 0-8 8-2/9-4/90 2-2/2- 1 (0-39) 1-9 33-3 [6] [6] [6] [2] [6] colonies encountered. Quite possibly, however, colonies less than 1 mm diameter may have been overlooked more commonly than the larger colonies. This potential bias in the collections has been compensated in part because where multiple encrusting cyclostomes are present on a single echinoid test or bivalve shell or fragment used in this study, each was assigned to species and used, whether or not the colonies had been marked by the original collectors. The Triassic species Reptomultispcirsa hybensis was collected by one of us (PDT), and all specimens encountered were kept. All substrates potentially encrusted by R. hybensis were collected, returned to the laboratory for cleaning and retained if encrusted. Collection of R. hybensis was therefore not selective. Data on colony size distributrions for species collected from the Oxford Clay 520 PALAEONTOLOGY, VOLUME 40 at Stanton Harcourt are not used, because specimens were screened in the field and only the more informative or well-preserved specimens kept, so that size bias may have been introduced. Material of two species, Hyporosopora dilatata and Plagioecia sp. 2, was collected specifically for this project from the Oxford/ Ampthill clays at Warboys Clay Pit, near Peterborough, Cambridgeshire. Between 300 and 400 specimens of Gryphaea and other oysters were collected without selectivity, brought back to the laboratory, and scanned for encrusting bryozoans. All encrusted bivalves were kept, and all colonies of the two bryozoan species were used in the study. Localities, stratigraphical horizons, and number of specimens of species used in this study are given in Table 1. It should be stressed that in most cases it has not been possible to establish the name for a species with complete confidence and precision because the systematics of Mesozoic encrusting cyclostomes is very poorly understood. Most of the species erected by 19th century authors, including d'Orbigny, Reuss and Roemer, were inadequately described and illustrated, and have never been revised using modern techniques such as SEM. In addition, the morphology of the gonozooid is crucial in generic identification and extremely valuable in distinguishing between species with similar autozooidal morphologies. However, gonozooidal characters are unknown in many species because gonozooids are lacking in the type specimens. Differences between species can be very subtle, and confident species determination is often only possible when type specimens are available for direct comparison. Colony sizes (maximum and minimum diameters, area, perimeter), diameters of zooidal apertures (ten measured per specimen), distances between centres of neighbouring zooidal apertures (ten measured per specimen), number of gonozooids, brood chamber sizes (maximum and minimum diameters, area), total area of brood chambers per colony, ancestrula to ooeciopore distances, and ooeciopore to colony edge distances were determined for each colony, using an ImageAnalyst system. Some quantitative characteristics of individual colonies were calculated, including 1. Average radius = (area/7r)° 5 2. Excess perimeter = perimeter — (2) (average radius) (rc) 3. Relative amount of excess perimeter = excess perimeter/((2) (radius) (7r)) INDIVIDUAL SPECIES Life history characteristics of individual species of Mesozoic encrusting cyclostomes are summarized in Table 2 (colony size attributes) and Table 3 (data on reproductive effort). Descriptions of the attributes of representative species for which the largest number of specimens is available are given below. EXPLANATION OF PLATE 1 Figs l^L Reptomultisparsa hybensis (Prantl, 1938); Triassic, Rhaetian, Hybe Beds; Hybe, Slovakia. 1-2, SNM 19779. 1. infertile, sub-circular colony with circumferential growing edge; x 10. 2, cluster of colonies including two small colonies, bottom left and bottom right, the former fan-shaped prior to the development of a circumferential growing edge, and the latter being overgrown by a larger colony; x 12. 3-4, SNM Z- 20655. 3, colony with lobe growing with a gonozooid towards the top left; x 10. 4, detail of the longitudinally elongate gonozooid; x 22. Figs 5-6. Hyporosopora dilatata (d’Orbigny, 1850); Jurassic, Callovian/Oxfordian, Oxford Clay; Stanton Harcourt, Oxfordshire. 5, BMNH D58679; regular colony with circumferential growing edge; abraded gonozooids are developed close to the growing edge lower left and lower right; x 7. 6, BMNH D59266; irregular colony abutting a serpulid and small cemented bivalves; a gonozooid is present in the lower left near the growing edge; x 5-6. All are back-scattered scanning electron micrographs of uncoated specimens, except figures 5 and 6 which are light photomicrographs. PLATE 1 McKINNEY and TAYLOR, Reptomultisparsa, Hyporosopora 522 PALAEONTOLOGY, VOLUME 40 average diameter (mm) B C D ancestrula to ooeciopore (mm) E % F ooeciopore to colony edge (mm) ooeciopore to colony edge (mm) text-fig. 1. Graphs summarizing quantitative data on Reptomultisparsa hybensis (Prantl, 1938). a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters), d, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. E, histogram of distances between ooeciopores and the nearest edge of the colony. f, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; horizontal alignment of points on this plot represents multiple values observed on a single colony; mean value for the colony is given as a open circle. Triassic species Reptomultisparsa hybensis. R. hybensis (Prantl, 1938) colonies grew as typically small, circular to somewhat irregularly shaped (Text-fig. 1a-b) sheets in which autozooidal apertures were isolated and quincuncially arranged. Minimum colony diameter averaged 83 per cent, of the maximum diameter. Most small but few large colonies were essentially circular (PI. 1, fig. 1); minimum diameter of the least circular colony was 56 per cent, of the maximum diameter, due to growth interference with a conspecific colony growing on the same substratum. In general, irregularity in shape was due at least in part to growth interference, which is clearly seen in several colonies that abut or overgrow but do not fuse with conspecifics (PI. 1, fig. 2). Subcolonies were not developed, although irregular lobes occur in some colonies (PI. 1, fig. 3). Coefficient of variation for mean diameter of colonies is 53 (for number (N) of colony measurements for this and other species, see Table 1). Plotted as percentage survivorship on a logarithmic axis (Text-fig. lc), average diameters of colonies show an increasing death rate up to approximately 5 mm, with somewhat irregular but near-constant death rate for larger colonies (Y = 23 1-3*10-° 198X). Minimum diameter of colonies with fully developed gonozooids was 4-2 mm (minimum area = 16-3 mm2), and gonozooids were present in four of the 27 colonies (15 per cent.) equal to or greater McKINNEY AND TAYLOR: BRYOZOAN LIFE HISTORIES 523 A average diameter (mm) B maximum diameter (mm) c average diameter (mm) D ancestrula to ooeciopore (mm) E ooeciopore to colony edge (mm) F colony edge (mm) text-fig. 2. Graphs summarizing quantitative data on Hyporosopora dilatata (d'Orbigny, 1 850). a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters), d, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. E, histogram of distances between ooeciopores and the nearest edge of the colony. f, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measurements, and open circles represent pooled data for each colony with two or more ooeciopores; regression, Y = — 5-04 + 52-75X. than 4-2 mm in minimum diameter (Text-fig. 1b). Minimum distance from ancestrula to ooeciopore was 2-0 mm, and mean distance 3 6 mm (Text-fig. Id; for number (N) of skeletal reproductive features of this and other species, see Table 3). Brood chambers (gonozooids) are much longer than broad, and most are close to the colony margin (PI. 1, fig. 4). The ooeciopore is an average of 0 9 mm from the outer edge of the developing zooids along the colony margin, with a range of 02-2-3 mm (Text-fig. 1e). There is no apparent correlation between colony area and distance between ooeciopore and colony margin (Text-fig. If; r = —0-094, p = 0-906, N = 4, using mean values from each colony with multiple gonozooids and observed values for colonies with a single gonozooid). Nine brood chambers occurred in one large colony and, with the exception of a single chamber located midway between the protoecium and colony margin, they occurred about 1 mm from the colony margin. The single chamber located mid-way was not included in the calculation of mean value of distance between ooeciopores and colony margin. No correlation between colony area and number of gonozooids in fertile colonies was seen in the very small sample of fertile colonies (r = 0-013, p = 0-987, N = 4). Minimum diameter of fertile colonies averaged 7-0 mm, while that of non-fertile colonies that had reached or exceeded the size of the smallest fertile colony averaged 5-3 mm. The difference is not 524 PALAEONTOLOGY, VOLUME 40 significant (Mann-Whitney U = 29, p — 0-246, N = 29), but only a small number of fertile colonies is present in the sample. Jurassic species Hyporosopora dilatata. Data presented here are for specimens collected from Warboys Clay Pit, Cambridgeshire. Colonies of H. dilatata (d'Orbigny, 1850) commonly grew to relatively large sizes (Text-fig. 2a). Early colony growth produced gradually widening fans (PI. 2, fig. 5) that within about 1 mm flared laterally so that both sides of the fan recurved and engulfed the proximal part of the ancestrula (protoecium). Subsequent growth produced sub-circular colonies (PI. 1, fig. 5) except that with increasing size colonies often became somewhat irregular in shape (Text-fig. 2b). Autozooidal apertures in H. dilatata are isolated and quincuncially arranged. Colonies included in this study encrust valves of the bivalve Gryphaea. They did not develop subcolonies, although a few have irregular lobes with a constricted region connecting the lobe(s) with the original part of the colony (PI. 1, fig. 6). Minimum diameter averages 75 per cent, of maximum diameter. Coefficient of variation for mean diameter of these colonies is 79. Plotted as percentage survivorship on a logarithmic axis, the average diameters of colonies shows a remarkably constant death rate with increasing size (Text- fig. 2c). Minimum diameter of colonies with fully developed gonozooids was 5-1 mm (minimum area was 22-0 mm2), and gonozooids were present in ten of the 43 colonies (23 per cent.) equal to or greater than 51 mm in diameter (Text-fig. 2b). Minimum distance from protoecium to ooeciopore is 2-2 mm, and the mean is 6 0 mm (Text-fig. 2d). There was no difference in mean minimum diameter of fertile colonies (10-4 mm) and of non-fertile colonies (10-1 mm) that had reached the 5-1 mm minimum diameter for fertile colonies (Mann-Whitney U = 163, p = 0-954, N = 43). The ooeciopore is located at about the midpoint of the dilated part of the gonozooid (PI. 2, fig. 6), averaging 1-8 mm from the outer edge of the developing zooids along the colony margin (Text- fig. 2e). There is moderate (r = 0-783, N = 10) but significant (p < 0-007) positive correlation between colony area and distance between ooeciopore and colony margin; although colonies grew only a short distance beyond completion of the brood chamber, the distance of continued growth was on average greater for larger colonies (Text-fig. 2f). This distance varied between 0-5 mm and 4-4 mm in the specimens available. Brood chambers are approximately equidimensional and are close to the colony margin. Fertile colonies had one to four brood chambers, and where two or more occur within a single colony, they are approximately equidistant from the edge of the colony. Among fertile colonies, there is a marginally significant correlation between colony area and number of fully developed gonozooids (r = 0-176, p = 0-094, N = 10). Plagioecia sp. 2. Except where otherwise specified, data presented for this species are for specimens collected from Warboys Clay Pit, Cambridgeshire. This species grew as typically small, essentially circular patches (Text-fig. 3a-b), formed by rapid expansion of an initially fan-shaped colony and coalesence of the two edges of the fan where they met and overgrew the early zooids of the colony (PI. 2, fig. 2). Autozooidal apertures are isolated and quincuncially arranged. Minimum diameter of colonies averages 83 per cent, of maximum diameter, and the majority, even the largest colonies (PI. 2, fig. 1), are near this ratio. Coefficient of variation for mean diameter of colonies is 42. Percentage survivorship plotted on a logarithmic scale shows increasing death rate up to about 4 mm average diameter, then an abrupt reduction in death rate followed by relatively slow increase in death rate for the larger colonies (Text-fig. 3c). Subcolonies (PI. 2, fig. 4) were produced in a small proportion of the colonies from Stanton Harcourt. They formed at the perimeter of the parent colony, originating from single zooids that functioned as pseudoancestrulae, from which a well-defined fan of zooids extended and gave rise to a radially expanding daughter colony. mckinney and taylor: bryozoan life histories 525 A average diameter (mm) B 7-t— 6- Cn 1 1 1 0 2.5 5 7.5 10 maximum diameter (mm) c average diameter (mm) D aiicestrula to ooeciopore (mm) E ooeciopore to colony edge (mm) F ooeciopore to colony edge (mm) text-fig. 3. Graphs summarizing quantitative data on Plagioecia sp. 2. a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters), d, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. E, histogram of distances between ooeciopores and the nearest edge of the colony, f, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measurements, and open circles represent pooled data for each colony with two or more ooeciopores; regression, Y = — I 1-19 + 36-54X. Minimum diameter of colonies with fully developed gonozooids was 2-5 mm (Text-fig. 3b; minimum area 5-8 mm2), and gonozooids were present in six of the 29 colonies (21 per cent.) equal to or greater than 2-5 mm in minimum diameter. Minimum distance from ancestrula to ooeciopore is 1-3 mm, and mean distance is 1-8 mm (Text-fig. 3d). Brood chambers are up to four times broader than long (PI. 2, fig. 3), and are elongated parallel to nearby colony margins. The ooeciopore is located on the distal side of the brood chamber, averaging 0-7 mm from the outer edge of the developing zooids along the colony margin (Text-fig. 3e). Distance between ooeciopore and colony margin is highly correlated with colony area (Text-fig. 3f; r = 0-880, p = 0-021, N = 6). Up to two brood chambers were noted per colony, although there was no correlation between colony area and number of brood chambers (r = 0164, p = 0-311) based on the small sample (N = 6) of fertile colonies from Cambridgeshire. Fertile colonies had a mean minimum diameter of 4-2 mm and mean area of 16-7 mm2 at time of death, and non-fertile colonies at least 2-5 mm in minimum diameter (the minimum observed diameter of fertile colonies) had a mean minimum diameter of 3-9 mm and mean area of 13-6 mm2. The differences in size are not significant (Mann-Whitney U test; minimum diameter, U = 57-0, P = 0-518; area, U = 55-0, p = 0-451 ; N = 29). 526 PALAEONTOLOGY, VOLUME 40 Cretaceous species Actinopora disticha. Colonies of A. disticha (von Hagenow, 1851) grew as small, essentially circular patches (Text-fig. 4a-b; PI. 3, figs 1, 3) in which autozooidal apertures are arranged in biserial, radiating fascicles. Specimens included in this study encrusted shells of inoceramid bivalves and echinoid tests. For solitary colonies and parental portions of compound colonies, minimum diameter averages 91-3 per cent, of maximum diameter. Coefficient of variation for mean diameter of these colonies that did not given rise to subcolonies is 49. Plotted as percentage survivorship on a logarithmic scale, the average diameters of solitary colonies and parent portions of compound colonies show an increasing death rate up to about 5 mm, with a few large outliers that have diameters greater than 5 mm (Text-fig. 4c). Subcolonies (PI. 3, fig. 2) extended from a marginal pseudoancestrular zooid in the parent colonies and were produced in 37 of 107 colonies (35 per cent.) equal to or greater than 1-2 mm in diameter, which was the smallest colony found to have produced a daughter subcolony. Mean diameter for production of subcolonies was approximately 3 0 mm (standard deviation = 0 98 mm, N = 37), with near-normal distribution except left-truncated at the point of the smallest parent colony (Text-fig. 4d). Usually only one daughter subcolony was produced. Rarely more than one first-generation daughter subcolony was budded from the parent, and only one definite and two or three ambiguous instances of budding of second-generation subcolonies were found. Colonies that did not produce subcolonies were usually circular, although a few of the largest were oval (Text-fig. 4b). Subcolonies were initially fan-shaped, with lateral walls of the fan formed of exterior wall. They then became essentially circular where the growth zone at the outer end of the fan flared and recurved laterally, joining and fusing over the proximal part of the cone. Continued growth of subcolonies occurred around the entire periphery. Local production of subcolonies around the colony perimeter therefore resulted in unequal colony diameters (Text-fig. 4b) and substantially lengthened the perimeter (Text-fig. 4e) such that where subcolonies are present there is a larger perimeter than would be necessary to enclose the total area of the colony were it circular (Text-fig. 4f). Minimum diameter of colonies with fully developed gonozooids was 2-2 mm (minimum area = 2 0 mm2), and gonozooids were present in 27 of the 74 colonies (35 per cent.) equal to or greater than 2-2 mm in diameter (Text-fig. 4g). Minimum distance from protoecium to ooeciopore was 0-6 mm, and the mean IT mm (Text-fig. 4h). Brood chambers are much broader than long, and are located close to and parallel with the colony margin. Fertile colonies had only a single brood chamber, or where multiple brood chambers were developed, they were in a single ring (PI. 3, fig. 3), having formed simultaneously around the colony perimeter. In a few colonies the entire circumference was occupied by two or three brood chambers joined at their lateral ends. The ooeciopore (PI. 3, fig. 4) is located on the distal side of the brood chamber, averaging 0-4 mm from the outer edge of the developing zooids along the colony margin (Text-fig. 4i). There is a moderate (r = 0-573; N = 27) but significant (p < 0 002) positive correlation between colony area and EXPLANATION OF PLATE 2 Figs l^L Plagioecia sp. 2; BMNH D59445; Jurassic, Callovian/Oxfordian, Oxford Clay; Stanton Harcourt, Oxfordshire. 1, part of a large colony with two rings of collapsed gonozooids; x 12. 2, central area of the same colony with the earliest zooids overgrown; x 40. 3, transversely elongate gonozooid with crushed frontal wall; x 60. 4, peripheral subcolony; x40. Figs 5-6. Hyporosopora dilatata (d’Orbigny, 1850). 5, BMNH D49316; Jurassic, Oxfordian, Oxford Clay; Warboys, Cambridgeshire; small fan-shaped colony with ancestrula (top right) not overgrown; x 14. 6, BMNH D58664; Jurassic, Callovian/Oxfordian, Oxford Clay; Stanton Harcourt, Oxfordshire; gonozooid with crushed frontal wall; x 30. All are back-scattered scanning electron micrographs of uncoated specimens. PLATE McKINNEY and TAYLOR, Plagioecia , Hyporosopora colony area (mm z) minimum diameter (mm) frequency frequency 528 PALAEONTOLOGY, VOLUME 40 A 80 t — 60- 40- ’ 20- : o f • 0 5 10 15 average diameter (mm) D B E C average diameter (mm) F G H ancestrula to ooeciopore (mm) I ooeciopore to colony edge (mm) J K ooeciopore to colony edge (mm) number of gonozooids text-fig. 4. For caption see opposite. mckinney and taylor: bryozoan life histories 529 distance between ooeciopore and colony margin; while colonies grew only a short distance beyond completion of the brood chamber (e.g. PI. 3, fig. 3), the distance of continued growth was on average greater for larger colonies (Text-fig. 4j). This distance varied between 0-05 mm and 0-95 mm in the specimens available. Up to four brood chambers were noted per colony, in most instances with relatively large fertile colonies having more brood chambers (Text-fig. 4k, r = 0 576, p 0 000). Fertile colonies had a mean minimum diameter of 4-4 mm and mean area of 18 0 mm2 at time of death, and nonfertile colonies that had minimum diameter of at least 2-2 mm (the minimum observed diameter of fertile colonies) had a mean minimum diameter of 3-5 mm and mean area of 111 mm2. The differences in size are significant (Mann-Whitney U test: minimum diameter, U - 333, p = 0-0007; area, U = 338, p = 0-012; N = 74). Among the colonies that were large enough to bear gonozooids, 28 had produced subcolonies, and of these nine had both subcolonies and gonozooids. There is no indication that either mode of reproduction is preferentially associated with or precluded by the other (X2 = 0-3667, p = 0-5434, N = 74). Discocavea irregularis. Colonies of D. irregularis (d’Orbigny, 1851) grew as typically small, circular colonies (Text-fig. 5a-b; PI. 3, fig. 5) in which autozooids radiate from the central region and are deflected obliquely upward, away from the substratum by new zooids budding basally around the colony perimeter of the mound-shaped colony. Because of the elongated cylindrical shape of the zooids and their divergence, the colony has a slight central depression atop the overall mound shape that is devoid of zooidal apertures. Throughout the remainder of the colony, zooidal apertures are slightly separated from one another and arranged in variably defined radial rows. For solitary colonies and parental portions of colonies that developed subcolonies, minimum diameter averages 91 per cent, of maximum diameter. Coefficient of variation for mean diameter of such colonies is 57. Plotted as percentage survivorship on a logarithmic scale, the average diameters of colonies show an increasing death rate, with minor inflections in the curve (Text-fig. 5c). Subcolonies were commonly produced, formed by a local lobe extending from the perimeter of the parent colony which then developed its own radial growth pattern, or by eruptive budding from the upper surface of the parent colony, with a new basal wall formed below the portion that spread laterally over the parent (PI. 3, fig. 5). The eruptive budding appears to have been centred on the inner portion of the peripheral ring of extending zooids of the colony rather than involving either the most marginal zooids or the inner portion of the colony where zooidal growth had slowed or ceased. Subcolonies were produced in 12 of 121 colonies (10 per cent.) equal to or greater than 1-7 mm in diameter, which was the smallest colony found that had produced a daughter subcolony (Text-fig. 5b). Four of the specimens with subcolonies had developed them by frontal eruptive budding. Mean diameter for production of subcolonies was 4-6 mm (standard deviation = 2-1 mm. text-fig. 4. Graphs summarizing quantitative data on Actinopora disticha (von Hagenow, 1851). A, histogram of average diameters of colonies. B, plot of minimum versus maximum colony diameters; filled circles represent colonies without subcolonies and open circles colonies with subcolonies, c, percentage survivorship based on size (average diameters), d, minimum diameter of parent portion of colony when first subcolony was produced. e, plot of colony perimeter length versus colony area ; filled circles represent colonies without subcolonies and open circles colonies with subcolonies, f, plot of excess perimeter above that required to enclose a circle of area equal to the colony, versus colony area; filled circles represent colonies without subcolonies and open circles colonies with subcolonies (regression for colonies with subcolonies, Y = 6-55(10OO51x)). G, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, h, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. i, histogram of distances between ooeciopores and the nearest edge of the colony. J, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measurements, and open circles represent pooled data for each colony with two or more ooeciopores; regression, Y = 619 + 31-80X. K, plot of number of brood chambers versus colony area; regression, Y = 6-75 + 6-83X. 530 PALAEONTOLOGY, VOLUME 40 N = 12), with a strongly right-skewed distribution (Text-fig. 5d). Multiple daughter subcolonies were commonly produced. Within the sample studied, brood chambers (PI. 3, fig. 6) were seen in only three colonies. They are inconspicuous, generally placed centrally, with a porous, interior-walled domal roof enclosing a space over the region where feeding zooids diverged forming the central depression of the colony. Lateral portions of the brood chamber generally extend for some distance along the surface of the colony, between the rows of autozooidal apertures. The position of the ooeciopore is unknown; none of the visible apertures are sufficiently differentiated to be identifiable as ooeciopores. Minimum diameter of fertile colonies ranged from 2-4 mm to 3T mm, encompassing the average minimum diameter of all colonies, and none of the three fertile colonies had generated subcolonies. Liripora complanata. L. complcmata (Roemer, 1840) grew as small, essentially circular colonies (Text-fig. 6a-b; PI. 4, figs 2-3) in which autozooidal apertures are arranged in uniserial, radiating fascicles. For solitary colonies and parent portions of compound colonies, minimum diameter averages 93 per cent, of maximum diameter. Coefficient of variation for mean diameter of such colonies is 25-3. Plotted as percentage survivorship on a logarithmic scale, the average diameters of colonies show an increasing death rate up to about 5 mm (Text-fig. 6c), with a decreased, near- constant death rate for colonies greater than 5 mm diameter. Subcolonies were commonly produced along the perimeter of the parent colony (PI. 4, figs 2-3). They are present in nine of 24 colonies (38 per cent.) equal to or greater than 3T mm in diameter, which was the smallest colony found that had produced a daughter subcolony. Mean diameter for production of subcolonies was 61 mm (standard deviation = F05 mm, N = 9), with a slightly left- skewed distribution (Text-fig. 6d). Usually only one daughter subcolony was produced per colony. Subcolonies, immediately following their initiation, developed a circular shape with holo- peripheral growth. Four parent colonies on specimen BMNH D46466 (PI. 4, figs 2-A) are distributed on the surface of an inoceramid as two pairs. Contact was made between colonies within each pair, growth stopped soon afterwards, and each parental colony then generated a subcolony. It seems possible that the production of subcolonies may have been stimulated either by contact between the parent colonies or by cessation of their growth. Within the sample studied, gonozooids were produced only in subcolonies, although a single specimen from another locality and horizon (BMNH D45165; B. mucronata Zone, Edward’s Pit, Mousehold, near Norwich, England), possibly belonging to the same species, has a gonozooid near the edge of the parent colony as well as one in a subcolony. Brood chambers (PI. 4, fig. 4) are recurved arcuate, broader than long, and are close to and parallel with the colony margin. The position of the ooeciopore is visible in only one specimen, where it is located on the distal side of the brood chamber (PI. 4, fig. 4). Minimum diameter of parent colonies with fertile subcolonies is 3-9 mm (Text-fig. 6e; minimum area 12-7 mm2), and gonozooids were present in seven of the 32 colonies (22 per cent.) equal to or EXPLANATION OF PLATE 3 Figs 1-4. Actinopora disticha (von Hagenow, 1851); Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Northfleet, Kent. 1-2, BMNH BZ 1838. 1, fertile colony (top left) and several smaller infertile colonies, some with peripheral subcolonies and some abraded; x 6. 2, peripheral subcolony; x43. 3-4, BMNH BZ 3232. 3, disc-shaped colony with two gonozooids forming a complete ring close to the growing edge; x 10. 4, small ooeciopore of a gonozooid located between radial rows of autozooidal apertures; x 60. Figs 5-6. Discocavea irregularis (d’Orbigny, 1851). 5, BMNH D45080; Cretaceous, Campanian, mucronata Zone, Upper Chalk; Earlham Lime Works, Norwich, Norfolk; colony with raised margins and a frontal subcolony; x 7. 6, BMNH BZ 1839; Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Northfleet, Kent; inconspicuous brood chamber made visible by partial loss of its roof; x45. All are back-scattered scanning electron micrographs of uncoated specimens. PLATE 3 McKINNEY and TAYLOR, Actinopora , Discocavea 532 PALAEONTOLOGY, VOLUME 40 A B C D minimum diameter of parent colony (mm) text-fig. 5. Graphs summarizing quantitative data on Discocavea irregularis (d’Orbigny, 1851). A, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles representing solitary colonies and open circles representing compound colonies, c, percentage survivorship based on size (average diameters). D, histogram of minimum diameters that parent colonies had attained when they produced a daughter subcolony. greater than 3-9 mm in diameter. Mean diameter of colonies with fully developed gonozooids is 7-9 mm, which is significantly different from the mean of 4-8 mm for non-fertile colonies equal to or greater in diameter than the 3-9 mm minimum for gonozooid development (Mann-Whitney U = 41, p = 0-034, N = 32). Based on the single preserved ooeciopore and presuming a similar position for ooeciopores of the abraded brood chambers, minimum distance from protoecium of the parent colony to ooeciopore is 1-8 mm, and the mean is 2-8 mm (Text-fig. 6f). Ooeciopores were located on average 05 mm from the outer edge of the developing zooids along the subcolony margin when the colony died (Text- fig. 6g). There is no correlation between colony size and distance between ooeciopore and subcolony margin (r = 0-750, p = 0-589, N = 7); instead, subcolonies, and parent colonies if they were still viable, grew only a short distance beyond completion of the brood chamber. This distance varied between 0-3 mm and TO mm in the specimens available. Only one gonozooid per colony was noted McKINNEY AND TAYLOR: BRYOZOAN LIFE HISTORIES 533 B C average diameter (mm) D minimum diameter with subcolonies E F ancestrula to ooeciopore (mm) text-fig. 6. Graphs summarizing quantitative data on Liripora complanata (Roemer, 1840). a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent colonies without subcolonies and open circles colonies with subcolonies present, c, percentage survivorship based on size (average diameters), d, histogram of minimum diameters that parent colonies had attained when they produced a daughter subcolony. E, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, f, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. G plot of distances between ooeciopores and the nearest edge of the colony, versus colony area. G ooeciopore to colony edge where they occurred in the sample studied, although in specimen BMNH D45161, which may belong to L. complanata , two gonozooids were present. ‘Mesonopora’ laguncula. " M\ lagancul a (Voigt, 1962) grew as small to moderately large, circular to rather irregularly shaped colonies (Text-fig. 7a-b; PI. 4, fig. 1 ) in which autozooidal apertures are isolated and quincuncially arranged. Zooids have their greatest external width about mid-way along the length, tapering both proximally and distally. Commonly, autozooids have their distal ends flexed to one side, giving a slightly sinuous external appearance to the zooids (cf. Serpen tipora: see Brood 1981). Colonies of this species resemble Walter’s (1989) concept of Mesonoporcu but mode of formation of the gonozooid differs from the type species and involves distal and proximal growth of the brood chamber from the maternal zooid. In addition, brood chambers are roughly equidimensional rather than being appreciably broader than long. The species probably belongs to an undescribed genus. 534 PALAEONTOLOGY, VOLUME 40 A B C D ancestrula to ooeciopore (mm) E ooeciopore to colony edge (mm) F ooeciopore to colony edge (mm) text-fig. 7. Graphs summarizing quantitative data on ‘ Mesonopora' laguncula (Voigt, 1962). a, histogram of average diameters of colonies. b. plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters), d, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. E, histogram of distances between ooeciopores and the nearest edge of the colony, f, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measure- ments, and open circles represent pooled data for each colony with 0 1 2 3 4 5 two or more ooeciopores. G, plot of number of brood chambers versus number of gonozooids colony area. 600 S 400- la o o 200- Minimum diameter of colonies averages 82 per cent, of their maximum diameter, and the variance is high. Even some large colonies are essentially circular, while maximum diameter of others is over twice minimum diameter. Growth interference is clearly seen in one colony that is slightly irregular in shape, but it is unclear why others became irregularly shaped. Coefficient of variation for mean diameter of colonies is 48. Plotted as percentage survivorship on a logarithmic scale, the average diameters of colonies show an increasing death rate up to about 16 mm (Text-fig. 7c), with three 'Methusalah’ colonies that reached diameters in excess of 20 mm. Colonies grew by a mixture of extension around the entire perimeter and local extension, but they did not produce subcolonies. Some colonies have pronounced growth lines that probably reflect variations in growth rate, including possible temporary growth cessation. Following this, however, resumption of higher growth rates occurred essentially around the entire perimeter, with neither peripheral pseudoancestrular groups producing subcolonies nor frontal budding of subcolonies. mckinney and taylor bryozoan life histories 535 A average diameter (mm) B maximum diameter (mm) c text-fig. 8. Graphs summarizing quantitative data on ' Mesonopora sp. a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters). D, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. E, histogram of distances between ooeciopores and the nearest edge of the colony, f, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measurements, and open circles represent pooled data for each colony with two or more ooeciopores; regression, Y = — 3-02 + 76-79X. G, plot of number of brood chambers versus colony area. G number of gonozooids Minimum diameter of colonies with fully developed gonozooids is 7-7 mm (Text-fig. 7b), minimum area 35-7 mm2, and gonozooids were present in ten of the 31 colonies (32 per cent.) equal to or greater than 7-7 mm in minimum diameter. Minimum distance from ancestrula to ooeciopore is 2-8 mm, and the mean distance is 4-5 mm (Text-fig. 7d). Brood chambers are approximately equidimensional, are located close to the colony margin (PI. 4, fig. 1), and where several occur within a single colony are scattered around much of the perimeter. The ooeciopore is located mid-way along the length of the brood chamber, averaging 1-5 mm from the outer edge of the developing zooids along the colony margin (Text-fig. 7e). Distance between the ooeciopore and colony margin varies from about 1 mm in some of the smaller colonies to as much as 2-7 mm in the largest fertile colony (Text-fig. 7f), but there is no correlation between colony area and this distance (r = 0-074, p = 0-840, N = 10). Up to four brood chambers were noted per colony, with no apparent correlation between colony area and number of brood chambers (Text-fig. 7g; r = 0-350, p = 0-321, N = 10). 536 PALAEONTOLOGY, VOLUME 40 Minimum diameter of fertile colonies averaged 10-5 mm, and their area averaged 92 -7 mm2. Non-fertile colonies that had reached or exceeded the size of the smallest fertile colony averaged 12-1 mm minimum diameter and had an average area of 105-9 mm2. The differences, however, are not significant (Mann-Whitney U test: minimum diameter, U = 84-5, p = 0-386; area, U = 81, p = 0-311; N = 31). ‘Mesonopora’ sp. This species grew as small to moderately large, circular to rather irregularly shaped colonies (Text-fig. 8a-b; PI. 4, fig. 5) in which autozooidal apertures are isolated and quincuncially arranged. As for ‘M. ’ laguncula , colonies of 'Mesonopora' sp. appear to fit within Walter’s (1989) concept of the genus, but mode of formation of the gonozooid differs from the type species, involving distal and proximal growth of the brood chamber from the maternal zooid and envelopment of the surrounding autozooids (PI. 4, fig. 6). The species probably belongs to the same undescribed genus as ‘M. ’ laguncula. Minimum diameter of colonies averages 85 per cent, their maximum diameter, and the majority are near this ratio. Even some large colonies are essentially circular, although the maximum diameter of others is over twice the minimum diameter. Irregularity in shape was due at least in part to growth interference, which is clearly seen in some colonies that abut other encrusting organisms (PI. 5, fig. 5) or that have lifted the growing edge apparently in an attempt to rise over some object or organism that was not preserved (PI. 4, fig. 5). Other irregularly shaped colonies show no evidence of interference from adjacent organisms or objects. Coefficient of variation for mean diameter of colonies is 56. Plotted as percentage survivorship on a logarithmic scale, the average diameters of solitary colonies and parent portions in compound colonies show a constant death rate up to about 15 mm (Text-fig. 8c). True subcolonies were not produced, although in a small percentage of specimens lobes extended locally well beyond the adjacent arrested or more slowly growing colony edge (PI. 5, fig. 5) and occasionally overgrew older parts of the colony (PI. 5, fig. 6). These lobes expanded only slightly once established. Minimum size of colonies with fully developed gonozooids is 2-7 mm (Text-fig. 8b; 6-4 mm2), and gonozooids were present in 41 of the 138 colonies (30 per cent.) equal to or greater than 2-7 mm in minimum diameter. Minimum distance from ancestrula to ooeciopore is IT mm, and the mean distance is 3-3 mm (Text-fig. 8d). Brood chambers (PI. 4, fig. 6) are broader than long, are close to and parallel with the colony margin, and may extend around much of the perimeter. The ooeciopore is located on the distal side of the brood chamber, averaging 0-8 mm from the outer edge of the developing zooids along the colony margin (Text-fig. 8e). There is a pronounced positive correlation between colony size and distance between ooeciopore and colony margin (r = 0-705, p < 0 001, N = 41). Ooeciopores are exactly on the colony margin in some of the smaller colonies, and are EXPLANATION OF PLATE 4 Fig. 1. ‘ Mesonopora' laguncula (Voigt, 1962); BMNH BZ 1837; Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Northfleet, Kent; part of a colony with two gonozooids visible as swellings centre left and top; x 10. Figs 2-4. Liripora complanata (Roemer, 1840); BMNH D46466; Cretaceous, Coniacian, cortestudinarium Zone, Upper Chalk; Seaford, Sussex. 2, two compound colonies, each with a peripheral subcolony; x 7-6. 3, peripheral subcolony (left) overgrowing the margin of the parent colony; x 15. 4, detail of the peripheral subcolony depicted in figure 3 showing the arcuate gonozooid with the ooeciopore immediately left of the hole in the roof ; x 4 1 Figs 5-6. ‘ Mesonopora ’ sp.; BMNH BZ 1844; Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Offham Hill, near Lewes, Sussex. 5, colony with two gonozooids (upper left and centre right) and a ridge (lower right) formed by overgrowth of an obstruction; x 12. 6, transversely elongate gonozooid enveloping several autozooids; x 57. All are back-scattered scanning electron micrographs of uncoated specimens. PLATE 4 McKINNEY and TAYLOR, ‘ Mesonopora\ Liripora 538 PALAEONTOLOGY, VOLUME 40 rarely more than 2 mm from the margin (Text-fig. 8f). Up to nine brood chambers were noted per colony, with relatively large colonies tending to have the higher number of brood chambers (Text-fig. 8g; r = 0-348, p = 0-026, N = 41). The colony with nine brood chambers had them arranged in two rows: an inner circular row of five, and an outer circular row of four, indicating that the colony had undergone two periods of reproduction, with death occurring shortly after completion of the outer circle (ooeciopore to growing edge averaging 0-32 mm for the outer row). Eleven of the colonies that had grown to 2-7 mm diameter or larger show evidence of interference within a short period before the colony died. This interference comprises abutment against other skeletized encrusters or development of elevated colony margins, although in most cases the objects or organisms encountered were not preserved. Over half (six of 11 ; 55 per cent.) of these colonies produced brood chambers shortly after contact with the obstruction. This proportion is marginally significantly higher than the fertile proportion (28 per cent.) of colonies of the same size range that show no preserved evidence of growth interference (X2 = 3-53, p = 0-06; N = 138). On one substratum, three approximately equal-sized colonies had fused and continued growing as a single chimaeric colony. Gonozooids were produced in each of the three original colonies at a distance from the ancestrula approximately equal to the distance from the ancestrula to the point of contact with the adjacent colonies, i.e. gonozooid production coincided with fusion. Fertile colonies had a mean minimum diameter of 7 0 mm and mean area of 53-4 mm2 at time of death, and nonfertile colonies that had minimum diameters of at least 2-7 mm (the minimum observed diameter of fertile colonies) had a mean minimum diameter of 5-8 mm and mean area of 34-1 mm2. These differences, however are not significant (Mann-Whitney U test: minimum diameter, U = 1678-5, p = 0-149; area, U = 1677, p = 0147; N = 41). Plagioecia? reniformis. P.l reniformis (Gregory, 1899) developed relatively large, circular to somewhat irregularly shaped colonies (Text-fig. 9a-b; PI. 5, fig. 1) in which autozooidal apertures are isolated and quincuncially arranged. Minimum diameter of colonies averages 76 per cent, of maximum diameter, and the majority are near this ratio. Even the largest colonies, while having generally more irregular outlines than smaller colonies, have an approximately 85 per cent, ratio of width to length. There is no preserved evidence for the cause of the moderate irregularities in colony outlines. Coefficient of variation for mean diameter of colonies is 41. Plotted as percentage survivorship on a logarithmic scale, the average diameters of colonies show an increasing death rate up to about 10 mm, at which point death rate decreased and then gradually increased (Text-fig. 9c). No colonies smaller than 3 mm diameter were found. True subcolonies were not produced, although in a small percentage of specimens, local lobes (PI. 5, fig. 3) extend well beyond the adjacent arrested or more slowly growing colony edge. The lobes have highly irregular shapes. Some lobes expanded only slightly once established but others EXPLANATION OF PLATE 5 Figs 1-4. Plagioecial reniformis (Gregory, 1899); Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Northfleet, Kent. 1--2, BMNH BZ 1836. 1, part of a colony with conspicuous growth checks and four visible gonozooids, two having broken roofs; x7-5. 2, gonozooid showing bulbous roof penetrated by a few autozooids and small distal ooeciopore (top centre); x 50. 3 — 4, BMNH BZ 1835. 3, broad lobe extending from the growing edge; x21. 4, growth irregularity comprising a spiral lobe of zooids growing across a truncated former growing edge; x 22. Figs 5-6. ‘ Mesonopora' sp. ; BMNH BZ 1051; Cretaceous, Santonian, coranguinum Zone, Upper Chalk; Northfleet, Kent. 5, lobate colony encrusting an echinoid test together with a stomatoporid bryozoan and a brachiopod (bottom right); x 11. 6, spiralling lobe overgrowing older parts of the colony; x26. All are back-scattered scanning electron micrographs of uncoated specimens. PLATE 5 McKINNEY and TAYLOR, Plagioecia ?, ‘ Mesonopora" PALAEONTOLOGY, VOLUME 40 540 average diameter (mm) D ancestmla to ooeciopore (mm) B 20-1— 0-1 1 1 1 1 0 5 10 15 20 25 maximum diameter (mm) E ooeciopore to colony edge (mm) F ooeciopore to colony edge (mm) G number of gonozooids text-fig. 9. Graphs summarizing quantitative data on Plagioecial reniformis (Gregory, 1899). a, histogram of average diameters of colonies, b, plot of minimum versus maximum colony diameters; filled circles represent infertile colonies and open circles colonies with complete gonozooids present, c, percentage survivorship based on size (average diameters), d, histogram of distances between point of origin of colonies (i.e. protoecium of ancestrula) and ooeciopores. e, histogram of distances between ooeciopores and the nearest edge of the colony. F, plot of distances between ooeciopores and the nearest edge of the colony, versus colony area; solid dots represent individual measurements, and open circles represent pooled data for each colony with two or more ooeciopores. G, plot of number of brood chambers versus colony area; regression, Y = 38-05 + 11-68X. expanded so much that the growth margin recurved (PI. 5, fig. 4) and locally overgrew or truncated the adjacent margin of the colony. Minimum diameter of colonies with fully developed gonozooids was 4-8 mm (Text-fig. 9b; minimum area 23 0 mm2), and gonozooids were present in 22 of the 40 colonies (55 per cent.) equal to or greater than 4-8 mm in minimum diameter. Minimum distance from ancestrula to ooeciopore is 1-9 mm. and the mean distance is 3-5 mm (Text-fig. 9d). Brood chambers (PI. 5, fig. 2) are up to twice as broad as long; the long axis is parallel with the nearby colony margin. The ooeciopore is located on the distal side of the brood chamber, averaging 0-9 mm from the outer edge of the developing zooids along the colony margin (Text-fig. 9e). There is no correlation between colony size and distance between ooeciopore and colony margin (Text-fig. 9f; r = 0-378, p = 0-091, N = 22). Up to eight brood chambers were noted per colony, with relatively large colonies tending to have the higher number of brood chambers (Text-fig. 9g; r = 0-639, p < 0-001, N = 22). Fertile colonies had a mean minimum diameter of 9-0 mm and mean area of 69-9 mm2 at time of mckinney and taylor: bryozoan life histories 541 death, and non-fertile colonies that had minimum diameter of at least 4-8 mm (the minimum observed diameter of fertile colonies) had a mean minimum diameter of 7 0 mm and mean area of 47-0 mm2. The differences in size are significant (Mann-Whitney U test: minimum diameter, U - 95, p = 0 005; area, U = 103, p = 0 010; N = 40). DISCUSSION Potential collection bias In order to estimate the degree of bias introduced by using museum collections for characterization of the life history attributes of the species used in this study, we duplicated measurements on two collections of the Jurassic species Hyporosopora dilatata and Plagioecia sp. 2. One suite of specimens for each of the two species was collected from Stanton Harcourt prior to this study, with strong field and laboratory selection for ‘well-preserved’, conspicuous specimens. The other was collected from the Warboys Clay Pit in Cambridgeshire by keeping all Gryphaea and other oysters encountered in the field, cleaning them in the laboratory, and then searching for all encrusting cyclostome colonies using hand lens and microscope. In addition, data on several colonies of Plagioecia sp. 2 found on a single museum specimen of Gryphaea dilatata from St Ives, Cambridgeshire, were combined with data on specimens from Warboys. Numerical evaluation of the suites of Hyporosopora dilatata specimens from Stanton Harcourt and Warboys yielded some similar and some different results, as can be seen in Tables 2 and 3 (pp. 518-519). The greatest difference is that the more intensive search for and retention of specimens from Warboys yielded a greater number of small colonies, which is reflected in smaller average diameters and area for Warboys H. dilatata and Plagioecia sp. 2, in comparison with those from Stanton Harcourt (Table 2). Distributions of colony size parameters are strongly different for the two suites of specimens, with near-normal distributions for the Stanton Harcourt H. dilatata sample (average diameter skewness 0-1997, kurtosis — 10133; area skewness 0-781, kurtosis —0-5219) and strongly right-skewed distributions for the Warboys sample (average diameter skewness 1-516, kurtosis 2-3755; area skewness 3-745, kurtosis 19-7818). These distributions differ significantly from one another (Kolgomorov Smirnov test, p < 0 001 for both measures). Because of the small sample size, statistical tests for differences between the two suites of Plagioecia sp. 2 were not made; the patterns of differences, however, appear to be similar to those for H. dilatata. Although we infer that collecting bias was entirely or largely the cause of the differences in size distributions of colonies of H. dilatata and Plagioecia sp. 2 between Stanton Harcourt and Warboys, we cannot rule out the possibility that somewhat different environments may have prevailed during deposition at the two sites that may have contributed to the differences. The different distributions of colony sizes resulted in strikingly different survivorship curves for Hyporosopora dilatata based on average colony diameter. Colonies that had been chosen while at Stanton Harcourt were all about 5 mm or more in diameter and produce a linear distribution of colony sizes (Text-fig. 10); in contrast, about half the specimens from Warboys were less than 5 mm in diameter. In addition, larger specimens were included in the Warboys collection than were in the Stanton Harcourt collection, possibly due to the largest colonies commonly being less pristine than smaller colonies and therefore being left behind at Stanton Harcourt. Consequently, the survivorship curve for Warboys H. dilatata is concave as plotted on arithmetic axes, in contrast to the linear, more steeply sloped curve for the Stanton Harcourt specimens. However, data on colony reproductive attributes are virtually identical for the two suites of specimens of H. dilatata , whereas it appears that fertile colonies of Plagioecia sp. 2 were collected preferentially at Stanton Harcourt (Table 3). We suspect that a lesser degree of selectivity, certainly no more than for specimens from Stanton Harcourt, was exercised by the collectors of the Cretaceous species used in this study. Consequently, the patterns determined here for all species are accepted by us as in general highly similar to the patterns that would characterize suites of specimens made without collector bias. The greatest differences would be in the range of colony sizes and the pattern of the survivorship curve; in 542 PALAEONTOLOGY, VOLUME 40 100 text-fig. 10. Comparison of percentage survivorship curves for Hyporosopora dilatata (d'Orbigny, 1850) based on size (average diameters), for a non-selective, thorough collection from Warboys and field- and laboratory-selected specimens from Stanton Harcourt. 0 5 10 15 20 25 30 average diameter (mm) addition, there may be a consistent slight overestimation of average colony size. Within-colony reproductive patterns as determined from the skeleton appear to be affected little if at all. Patterns of life history characteristics Survivorship curves. ‘Survivorship’ as used in this paper is based entirely on the pattern of size- frequency distributions of available colonies within a species. Size-frequency distributions of population samples are influenced by several intrinsic attributes of a species, including number of recruits per class, seasonality of recruitment, rate of growth of individuals, coefficient of variation of growth rate, mortality rate, and seasonal interruptions in growth that are due either to adverse conditions or to reproduction (Craig and Oertel 1966). Fossil assemblages are almost always time- averaged, and they are necessarily time-averaged where specimens are taken from multiple beds as in this study. Therefore recruits per class, seasonality of recruitment, and seasonal interruptions in growth cannot be analysed for the assemblages reported here, although growth interruptions can be recognized in individual specimens of some species such as ‘ Mesonopora' laguncula and Plagioecial reniformis (PI. 5, fig. 1). Setting aside possible taphonomic and collector biases to size distributions (see discussions above), the size-frequency distributions of the fossil assemblages and the size-based survivorship curves derived from these distributions are primarily influenced by growth rate and by mortality rate. Other than uncommon partial overgrowth, none of the encrusting cyclostome species in this study show any indication of negative growth (shrinkage), which may be caused by partial overgrowth, partial predation, or fission. Several studies have shown that growth rate, onset of reproduction, reproductive output, and mortality are influenced by size rather than by age in clonal organisms (e.g. Hughes and Jackson 1980). Although within each species in this study size is presumed to correlate closely with age, size is not used as a proxy for age but rather as a measure of colony development against which mortality rate and other aspects of life history are compared. Whether based on age units or size units, mortality rate patterns (Text-fig. 1 1) generally fall into three groups: (1) high mortality in infancy, decreasing to low death rate; (2) constant mortality, in which the same proportion of the population dies per unit time/size; and (3) low initial mortality followed by an increase to high mortality rate (Craig and Oertel 1966). A special case of the latter is a constant number of deaths per unit time/size. Each of the three species in which unselective collections were analysed shows a somewhat different survivorship curve. The overall curve for Reptomultisparsa hybensis (Text-fig. lc) is mckinney and taylor: bryozoan life histories 543 text-fig. 11. Survivorship curves, a, high initial mortality followed by decreased mortality rate, b, constant mortality rate, c, low initial mortality followed by abruptly increased mortality rate, d, low initial mortality with constantly increased mortality rate due to constant number dying for each size unit. somewhat irregular but is overall essentially linear or slightly convex, indicating a constant or slightly increasing mortality rate with colony size. The first portion of the curve, up to about 5 mm, is distinctly convex if plotted with survivorship on a logarithmic scale; plotted on a linear scale, this portion of the curve is straight, indicating that deaths were constant per unit size. The survivorship of Hyporosopora dilatata (Text-fig. 2c) is linear and indicates a constant probability of death per unit size. In contrast with the first two species, the survivorship curve for Plagioecia sp. 2 (Text-fig. 3c) is convex up to 4 mm average diameter, indicating increased mortality rate, and linear above 4 mm average diameter, indicating constant mortality rate for such larger colonies. Although the survivorship curves of the other species are potentially influenced by collecting bias, the range of shapes found roughly corresponds with the range of shapes of curves for the three unselectively sampled species described above. Actinopora disticha (Text-fig. 4c), ‘ Mesonopora ' laguncula (Text-fig. 7c), and Plagioecia^. reniformis (Text-fig. 9c) show constantly increasing mortality rates for most of the population samples, reflecting a constant number of deaths per unit size, until a certain size threshold was reached, at which point mortality rate declined for the few large colonies that exceeded this size. Discocavea irregularis (Text-fig. 5c) and ‘ Mesonopora ’ sp. (Text-fig. 8c) have survivorship curves essentially similar to that of Reptomultisparsa hybensis (Text-fig. lc). The survivorship curve for Liripora complanata (Text-fig. 6c) is more complex but overall nearly straight or slightly convex. About 60 per cent, of the L. complanata colonies died between diameters of 3 mm and 5 mm, but those that passed the 5 mm threshold could reach much larger sizes. None of the species studied shows any evidence of reaching a maximum size beyond which growth could not occur. Even species such as Actinopora disticha and 'Mesonopora' laguncula , and Plagioecia ? reniformis , for which pronounced convex survivorship curves show that mortality rate increased for the majority of the size distribution of colonies, include several specimens that ‘escaped’ to grow into substantially larger colonies. These larger, ‘Methusalah’ colonies may have grown anomolously rapidly during the normal life span of colonies of the species, or they may represent colonies that survived beyond the normal life span, perhaps to continue living into or through the following growth season. The survivorship curves for Mesozoic cyclostomes, including those potentially influenced by collecting bias, are similar to curves found in some Eocene and Recent encrusting cyclostomes (McKinney el al. 1996; G. M. Galloway, pers comm.). Some of these post-Mesozoic encrusting cyclostomes exhibit increasing mortality rate generated by a constant number of deaths per unit size, while others are characterized by constant mortality rate. The post-Mesozoic species reached. 544 PALAEONTOLOGY, VOLUME 40 on average, smaller colony diameters than did the Mesozoic species, but also include a small number of relatively large, ‘Methusalah’ colonies. These mortality patterns contrast with Hakansson’s (1976) findings in a study of two free-living Cretaceous cheilostome bryozoans, Stichopora pentasticha (von Hagenow) and Lunulites mitra von Hagenow. Both species evidently suffered high juvenile mortality rates and probably decreasing growth rates, resulting in population structures similar to those seen in non-clonal benthic marine invertebrates. Colony sizes and shapes. Equidimensionality of colonies, as expressed by the ratio of minimum to maximum colony diameters, was predicted to decrease with increasing colony size and to be positively correlated with the width to length ratio of brood chambers. In other words, we expected that in species that usually grew to larger sizes the colonies would be more irregular in shape than those that usually grew to smaller sizes, and that those species in which brood chambers are very broad, essentially parallel with colony margins, would have grown to more circular shapes than those with equidimensional or longitudinally elongated brood chambers. However, although the ratio of minimum to maximum colony diameters is negatively correlated with all measures of colony size (minimum diameter, maximum diameter, area), none are significant (Table 4). Also contrary to prediction, ratio of mean values for minimum and maximum diameters for each species was not correlated with shape of brood chambers (Table 4). Table 4 lists correlation values based on the ratios of mean values of minimum and maximum diameter for each species; the same patterns of non-significance were obtained when ratios based on median values were used. If characterized by median values rather than mean values, 12 of the 15 species have more nearly equidimensional colonies, indicated by higher ratios in Table 2. This is statistically different from the null hypothesis of neither being preferentially larger (X2 = 5-40, p = 0-020, df = 1, N = 15) and suggests that extreme values for individual colonies within most species have a measureable effect on the ratios, because the degree of asymmetry of individual colonies would affect the mean values more than the median values. Twelve of the species do indeed have larger mean colony diameters than median colony diameters (Table 2). However, a test for correspondence of low ratio of mean values for colony diameters with larger values for colony means rather than medians fails (X2 = 1-67, p = 0-194. df = 1, N = 15). Apparently some other source of difference in colony equidimensionality is also involved, perhaps the high asymmetry of very young colonies (e.g. PI. 1, fig. 2; PI. 2, fig. 5) before they attain a disc-shape with a circumferential growing edge. Zooid sizes. We used two measures to characterize size of zooids for each species, aperture diameter and centre-to-centre spacing between neighbouring zooidal apertures. Both of these measures are correlated with zooidal tentacle length, diameter of lophophores when in the feeding position, and with diameter of mouth (Winston 198 1 z/ ; McKinney and Jackson 1989) in living cyclostomes. They are therefore related to strength of feeding currents and to potential rate of nutrient intake and growth rate (Winston 1977; McKinney 1993). Aperture diameter is significantly correlated with zooidal spacing but is not correlated at p < 0-05 with any other characteristics (Table 4). Zooidal spacing correlates significantly with all measures of colony size and with presence of subcolonies, and correlates at p < 0-001 with distance between protoecium and ooeciopore, ooeciopore to adjacent colony margin, and minimum size at reproduction. These latter three characteristics are all related to colony size measures and so part of their correlation with zooidal spacing derives from that. However, each of them is much more highly correlated with zooidal spacing than are measures of colony size, so that zooidal size itself must have a close relationship to size at reproduction and survival after reproduction in cyclostomes. Subcolonies. Subcolonies are absent in most Triassic and Jurassic species included in this study, although irregular, lobate extensions from the colony perimeter do occur in Reptomultisparsa table 4. Kendall’s rank order correlations for life history characteristics of Mesozoic encrusting Cyclostomata. The correlation coefficients are followed by p values, which are enclosed in brackets. Coefficients for which p > 0 05 are printed in boldface. N = 14 for all correlations involving brood chamber shape and placement of the ooeciopore; for other correlations, N = 15. mckinney and taylor: bryozoan life histories 545 00 c- r- — o © o cm io (N o o oo oo O oo (N ^ © © © © I w I w o I o CJ Tt CM o OOn t O (N (N O O l/M O OO — © © ON CM ON CM © © CM OO r- CM oo CM © «— ' CM © c~ © oo © © © © © © 1 1 1 CM oo 1— H CM ON — On c- i— < c~ CM ON M- r- i—> ON CM Cl CJ CM © CM CM © © © © © © © © 1 1 W © ON CM CM CM i— i CM OO © ON © r-» ON X © r oo © © © — CM Cl — Cl c © © © © © © © © © © © 1 1 W NO Mf N© © CM © © oo ON Cl ON »Ti © 00 © © OO © i—i 1— < r- r— rC © NO © CM — © CM 1—1 CM © r- © © © © © © © © © © © © © 1 1 o CM — 00 3 CM © OO r © © C 1 oo — CM re © o IT) © r- © © © CM © © CM © 00 o © ci — < © © CM CM — i 00 © © © © © © © © 3. © © © © © © 1 w 1 _ _ r __ _ _ ' — -> — NO s NO o o o CM © © © CM — — Cl OO © © ON o 00 o ON © CM i— < Cl c~ Cl © 0 i—i r- o c o NO © cm © CJ CM CM CJ CM <— 1 UTi © © © © © © © © © © © © © © © © c> 1 1 NO NO c~ Co ON '"t — oo ON OO oo © CM — -H oo CM cm r NO cm cm ON Mf © ON — — ON — CM — < CM cm CM i — © OO © © CM — CM © © © © © © © © © © © © © © © © 1 1 1 1 1 1 CM oo 00 o 00 o © © — -H © CM ON c- © ITi CM r- o 1 Ct o r- © r- © Mj- — > ON *— < Mf OO ON — r- o r- o NO © cm © C1 CM CM CM © © © © © © © © © 1 © © © © 1 © © © © © (N CM o 00 o l/M 3 c- CM OO i—i c~ Cl OO © © © r- o i/m c « oo CM © CM oo CM r- oo i—i r- o o © CM Cl Cl — < Cl IT> © © © © © © © © © © © © © © © © © G *3 g D — O 3 i-H o 3 d- E: a ■5 d a .2 o 03 QJ U-, C/5 1> CJ O C/5 o 8 o % o w IX o 'o «D O § © < < o PM c o 'O O flo 1) OJ d> ^ r- 2 2 — § aj -o O 2 ° X 03 3 a* t: g & -3 Cu G o3 a 03 © ’o o N 546 PALAEONTOLOGY, VOLUME 40 hybensis (PI. 1, fig. 3) and Hyporosopora dilatata (PI. 1, fig. 6). We do not consider such lobate outgrowths to be subcolonies because they originate from variably broad portions of the colony perimeter and there is no conspicuous and consistent differentiation between them and the original part of the colony from which they extend (e.g. no secondary zone of astogenetic change). One Oxfordian species, Plagioecia sp. 2, gave rise to peripheral subcolonies from single zooids that functioned as pseudoancestrulae from which a well-defined fan of zooids formed a radially expanding, secondary growth centre readily recognizable as a subcolony (PI. 2, fig. 4). A similar pattern of subcolony origination (PI. 3, fig. 2) is seen in the Late Cretaceous species Actinopora disticha. Alternative patterns of subcolony formation were present in other Late Cretaceous species. Discocavea irrgeularis had both local proliferation of budding from a small group of peripheral zooids that gave rise to satellite daughter subcolonies with central maculae resembling those in the parent colony, and also eruptive budding from within the central portion of the colony to form one or more basal wall-bounded, ‘stacked’ frontal subcolonies above the parent colony (PI. 3, fig. 5). Discocavea irregularis developed raised colony margins (PI. 3, fig. 5) much more commonly than did other species included in this study. Some modern cyclostomes also elevate the colony margin upon approach or contact with competitiors, which serves to defer overgrowth by encroaching encrusters (Stebbing 1973; McKinney 1992). Among eight species of northern Adriatic encrusting cyclostomes, several species occasionally develop raised colony margins, but Plagioecia patina does so commonly (McKinney 1992). In addition, P. patina often develops frontally budded daughter subcolonies (McKinney 1992, fig. 2a-c). The correspondence between propensity to elevate the growing margin of the parent colony and frontal budding of daughter subcolonies may be a general pattern in cyclostomes. For those species with the capacity to initiate daughter colonies above the parent colony surface, elevation of the growing margin may stop or delay overgrowth by an approaching competitor long enough for the elevated daughter colony to become established. An analogous situation exists in some Recent cheilostome bryozoans: Antropora tincta is able to overgrow the cheilostome Onychocella alula because, although A. tincta has smaller zooids than O. alula , A. tincta can frontally bud new layers whereas O. alula cannot. Where A. tincta can block lateral growth of O. alula , it then buds frontally and spreads laterally over the top of O. alula (Buss 1981). This is but one example of how topographical advantage can influence the outcome of overgrowth competition (Walters and Wethey 1986). The topographically higher position of the daughter subcolonies in D. irregularis , P. patina and other cyclostomes with the same potential for frontal budding of daughter subcolonies may enable them to survive after overgrowth of the parent colony and may even serve as a position from which to spread over a competitor that has covered the lower, parent colony. Therefore, elevated margins of parent colonies and production of frontally budded daughter subcolonies probably extend the life span of the colony as a whole, and the two characteristics together constitute an important life history trait. Subcolonies provide new regions of growth and therefore increase the area of those colonies in which they are developed. However, formation of subcolonies occurred predominantly in species with relatively small colony sizes among the species included in the study; generally, species that produced colonies characterized by a non-subdivided encrusting sheet grew to larger sizes (Mann- Whitney U = 700, p = 0027, N = 15; identical results based on minimum diameter, maximum diameter, and colony area). The only other life history attributes listed in Table 4 to which presence of subcolonies is related, are the distance from the point of origin of the colony to the closest ooeciopore (Mann-Whitney U = 2-00, p = 0-01 1, N = 14) and the mean distance from the point of origin of the colony to all ooeciopores in the colony (Mann-Whitney U = 3 00, p = 0-016, N = 14), both of which are highly correlated with colony size (Table 4). Fission of encrusting bryozoan colonies, typically caused by local abrasive grazing or by overgrowth separating two or more parts of a colony, is common for some species that live on hard substrata at the present day (Jackson and Winston 1981 ; Vail and Wass 1981u, 1981fi; Winston and Jackson 1984). None of the species studied here was seen to reproduce asexually by fission, even those in which the existence of peripheral subcolonies derived from single-zooid ‘pseudoancestrulae’ would seem to offer potential for fragmentation. The absence of fragmentation among the mckinney and taylor: bryozoan life histories 547 specimens studied was possibly due to some combination of relatively small colony size and perhaps less pervasive grazing damage in the Mesozoic than at the present day (Vermeij 1987). Si:e at sexual reproduction. Sexually reproductive male zooids are skeletally indistinguishable from non-reproductive autozooids (feeding zooids) in cyclostome bryozoans (Harmer 1896; Borg 1926; Ryland 1970; Silen 1972, 1977), and in a large proportion of bryozoan species the normal feeding zooids produce sperm from early in their individual life cycles (Mawatari 1975). Therefore, the onset of sperm production and release cannot be determined from fossils or the skeletal remains of Recent cyclostomes. In contrast, brood chambers produced by female zooids in cyclostomes have characteristic skeletal morphologies (e.g. Silen 1977; Strom 1977; Schafer 1991) which are readily recognizable in fossil and Recent skeletal remains. In most of the studied species brood chambers comprise gonozooids that are broader and more bulbous than the autozooids in the same colony (e.g. PI. 2, figs 3, 6; PI. 3, fig. 4; PI. 4, fig. 6; PI. 5, fig. 2). Such brood chambers usually have a roof of densely pseudoporous exterior wall. However, in Discocavea irregularis the brood chambers are roofed by interior walls and are much less conspicuous (PI. 3, fig. 6). Throughout this paper, the terms ‘fertile colony’ and ‘reproduction’ refer to colonies in which are present brood chambers, complete with the ooeciopore through which the larvae were released. There is a strong, positive correlation between the colony size typically attained by species included in this study and the size at which they first reproduced and at which the major reproductive effort was made (Table 4; see ‘Ancestrula to 1st ooeciopore' and ‘x ancestrula to ooeciopore’). Both of these are reasonable outcomes: colonies in species that reached larger sizes had a greater range of distances from the point of colony origin at which to form gonozooids. Similar results in correlation of average colony size and average size of reproductive colonies have recently been documented for three co-occurring species of the encrusting cheilostome genus Stylopoma (Herrera et al. 1996). It is conceivable that colonies in some species, even though commonly reaching large sizes, may be genetically programmed consistently to initiate reproduction at small colony size. Histograms of distances from ancestrulae to ooeciopores show a variety of distribution patterns. The most tightly constrained pattern is seen in Actinopora disticha , in which almost all brood chambers developed close to the point of origin of the colonies, with the ooeciopores closely grouped approximately 1 mm away (Text-fig. 4h). Plagioecial reniformis also tended to reproduce at a fairly uniform distance from the colony origin, with most brood chambers having been completed between 2 mm and 4 mm from the origin. These patterns of limited range in size of colonies at time of reproduction seem to indicate that reproductive maturity for A. disticha and P.l reniformis was a constrained part of the astogenetic developmental programme characteristic of the species, even though only a small proportion of colonies in each of the species reproduced. However, none of the Mesozoic cyclostome species has such a highly regulated onset of female reproduction as seen in the living cheilostome Celleporella hyalina , in which reproduction begins at about the 57-zooid stage, regardless of the rate at which the colony grew to that stage (Cancino and Hughes 1987). In contrast with the patterns of distribution of gonozooids found in Actinopora disticha and Plagioecial reniformis , ooeciopores of the highly elongate brood chambers (PI. 1, fig. 4) of Reptomultisparsa hybensis are relatively uniformly scattered across a wide range of distances (2 mm to 6 mm) from the point of colony origin (Text-fig. Id). Distribution patterns of distances between colony origins and ooeciopores in other species in this study tend to be closer to that of R. hybensis than to the other end-member patterns. The highly deterministic pattern of small range in distribution of colony origin to ooeciopore distances in Actinopora disticha is consistent with a ‘determinate’ growth pattern in which reproduction occurs at small colony size, after a short period of growth, largely controlled by an intrinsic, relatively rigid programme of growth. The broad range of distances between the point of colony origin and ooeciopores in Reptomultisparsa hybensis suggests that in this species reproduction is largely stimulated by some environmental signal. For some colonies in several species (Hvporosopora dilatata, Liripora complanata , ‘ Mesonopora ' laguncula , ‘ Mesonopora ’ sp., Plagioecia 548 PALAEONTOLOGY, VOLUME 40 sp. 2) the probable environmental stimulus can be identified: approach to or contact with an obstruction, commonly another skeletized organism, either a conspecific colony or a representative of a different species. For most species, there was no statistical difference in size of fertile colonies and size of non-fertile colonies that had reached or passed the size of the smallest fertile colony. For three species (. Actinopora disticha , Liripora complanata , Plagioecia'! reniformis ), however, fertile colonies were on average significantly larger than non-fertile colonies that had reached or passed the size of the smallest fertile colony. Therefore, although reproduction was followed by only slightly more growth before death of the colony, reproduction did not result in colonies of smaller-than-average size. On the contrary for the three species cited above, reproduction was more characteristic of larger colonies. Most of the species in this study, therefore, have a reproductive pattern that to some degree resembles that previously documented for several cheilostome bryozoans, especially Electra posidoniae (see Silen 1966) and species of Membranipora (Flarvell and Grosberg 1988; Cancino et al. 1991 ; Karande and Udhayakumar 1992). Crowding of M. membranacea by conspecifics triggers the onset of reproduction across a broad range of colony sizes (3-30 mm diameter; Harvell and Grosberg 1988). In Membranipora isabelleana (see Cancino et al. 1991), onset of both male and female reproduction is highly flexible in size and age, with early reproduction induced by conspecific crowding and characterized by sexual reproduction at the crowded edge. Where uncrowded, M. isabelleana eventually reproduced sexually in centres of colonies. Apparently in the thiee cyclostome species ( Actinopora disticha , Liripora complanata , Plagioecia! reniformis ) in which fertile colonies were on average larger than infertile colonies, larger colonies tended to become reproductive more readily than younger colonies even in the absence of contact or crowding, as seen by Cancino et al. (1991) in M. isabelleana. The lack of statistical difference in size of fertile and nonfertile colonies for other Mesozoic cyclostomes implies that they had not reached a size at which reproduction is increasingly likely. Where there is no skeletal evidence of crowding, the stimulus (if extrinsic) for onset of reproduction cannot be determined. Among the possibilities are contact or imminent contact with a non-preserved neighbour, change in temperature (Dudley 1973), plankton bloom (Gordon 1970), grazing (Harvell and Grosberg 1988), and fine-scale variation in the physical environment (Keough 1989). Recent evidence has shown that turbulent diffusion can reduce sperm availability and influence female reproductive success in free-spawning marine animals (Levitan and Petersen 1995). Therefore, sperm availability is another factor that could potentially determine the occurrence of female reproduction in cyclostomes (Ryland 1996) assuming that brood chambers develop only after eggs have been successfully fertilized, an idea that requires testing using living colonies. Colony survival after reproduction. With the exception of Discocavea irregularis , all of the species included in this study have brood chambers that developed at or very close to the distal growing edge of the colony (e.g. PI. 1, fig. 5; PI. 3, fig. 3; PI. 4, figs 1, 5). Continued growth of the colony generated an ever increasing growth increment beyond the completed brood chambers. In most species, where multiple brood chambers occurred within a single colony, they were formed almost simultaneously, as indicated by a relatively uniform distance between each of the brood chambers within the colony and the adjacent margin to which the colony had grown at death (e.g. PI. 1, fig. 5; PI. 2, fig. 1 ; PI. 3, fig. 3; PI. 4, figs 1. 5; PI. 5, fig. 1). In all species apart from D. irregularis , regardless of how large a colony had grown before formation of brood chambers, growth in most instances continued for only a short distance beyond the completed brood chambers (Table 3). Species characterized by larger colonies tended to grow for greater distances following completion of brood chambers (Text-fig. 12; Table 4). In addition, for Actinopora disticha , Hvporosopora dilatcita , ‘ Mesonopora ’ sp., Plagioecia sp. I and Plagioecia sp. 2, there is significant (P <0-01) positive correlation between size of colonies within the species and distance beyond ooeciopores to which the colonies continued to grow. mckinney and taylor bryozoan life histories 549 text-fig. 12. Plot of species averages (means) of distance between ooeciopores and adjacent colony edges, versus species averages (means) of colony areas. Filled circles represent species with narrow brood chambers (width : length ratios < 1-5; regression, Y = 0T71X0516), and open circ- les represent species with broad brood chambers (width: length ratios >1-5; regression, Y = 0T43X0450). text-fig. 13. Plot of standard deviations of distance between ooeciopores and adjacent colony edges, versus species averages (means) of brood chamber width: length ratios. Each point on the graph represents a value from a single colony in which there are three or more gonozooids. standard deviation, ooeciopore to colony edge (mm) In species with broad brood chambers, especially where the brood chambers followed the curvature of the colony margin as in Actinopora disticha, a conspicuous ring of brood chambers developed that in some instances extended the full 360° around the colony perimeter (PI. 3, fig. 3). Colonies of such species appear to have been semelparous, reproducing over a short period and dying soon after this time of high expenditure of energy. Semelparous reproduction followed by death of the colony is common for cyclostome species living at the present day and has been documented for Crisidia cornuta (see Eggleston 1972), Disporella ovoidea (see Winston 1985), D. plumosa Winston and Hakansson, 1986, and Lichenopora verrucaria (see Harmer 1896). Colonies in Mesozoic species with equidimensional or elongate brood chambers (e.g. PI. 1, fig. 4) commonly do not have such conspicuous alignment of brood chambers and give the appearance of having reproduced over a longer period and to have survived longer following production of the brood chambers. Less precise alignment of equidimensional or elongate brood chambers around the 550 PALAEONTOLOGY, VOLUME 40 periphery of colonies is not supported by significant correlation of each species’ standard deviation for mean distance from ooeciopores to colony edge, compared with the species’ brood chamber width: length ratio (Kendall’s rank order correlation, tau — —0 015, p = 0-094, N = 12). However, if within-colony means of the distance between ooeciopores and the adjacent colony edge are compared with the species’ brood chamber width: length ratios, there is a highly significant negative correlation (Text-fig. 13; Kendall’s rank order correlation, tau = —0-254, p = 0 0035, N = 62, using individual colonies with three or more brood chambers from all species in the study). That is, the broader the brood chambers with respect to their length, the more precisely they are arranged around the colony perimeter where three or more occur within a colony. Species with small colonies grew only slightly more after completion of brood chambers, regardless of the shape of the brood chambers. However, species with larger colonies do fall into two groups; those with equidimensional or elongate brood chambers continued to grow further than did those with broad brood chambers (Text-fig. 12; Table 4). No species with broad brood chambers averaged more than IT mm growth beyond the positions at which brood chambers formed, even those with colony areas averaging up to 100 mm2, whereas for species with equidimensional or elongate brood chambers, there is a uniform correspondence between average colony area and average distance between ooeciopores and colony edge (Text-fig. 12). The proximity of brood chambers to the colony margin at time of death, and their occurrence in a single near-marginal ring where multiple brood chambers occur, indicate that in almost all instances in these exterior-walled species there was a single period of reproduction, followed shortly by death. Formation of a completed brood chamber was probably the less energetically costly part of reproduction, exceeded by the subsequent nourishment of the embryos which are rich in protein- and lipid yolk (Dyrynda and King 1983). This period of nourishment of the developing embryos would have corresponded with the time during which a colony continued to grow the short distance between brood chamber completion and the final colony margin. It is unusual to find colonies that have brood chambers in two distinctly different positions, one inner and the other outer, but this occurred in single colonies of some species (e.g. ‘ Mesonopora sp., Hyporosopora sp., Reptomultisparsa hvbensis , Plagioecia sp. 2: PI. 2, fig. 1), indicating two distinct periods of reproduction (iteroparity). This occasional occurrence of two distinct reproductive periods in otherwise semelparous cyclostome species is known in some living species. Lichenopora verrucaria and Disporella ovoidea are small, encrusting cyclostomes with overall colony morphology much like that of Discocavea irregularis in this study. In colonies of L. verrucaria that live into a second year, it is common for subsequent broods of embryos again to be produced, occupying the original centrally located brood chamber (Harmer 1896). D. ovoidea colonies in Jamaica have a ’pseudo-solitary’ ecology, with rapid growth to finite adult size, a single central excurrent outlet for filtered water, and one colony brood chamber; most die after a single period of sexual reproduction, but a few colonies survive and reproduce again after a few months (Winston 1985). It is impossible to determine the relationship between sexual reproduction and death of the colony for Discocavea irregularis because the potential for brood chambers to develop proximally of the growing edge in this free-walled cyclostome does not allow determination from skeletal evidence of how much further a colony grew after reaching sexual maturity. Only three out of 146 specimens (2 per cent.) showed clear brood chambers. This contrasts strongly with Harmer’s (1896) observations for the morphologically similar Lichenopora verrucaria in which all colonies became reproductive upon reaching approximately 1 mm diameter, after which growth continued for a few more millimetres. Reproductive effort. The proportion of colonies that became fertile upon reaching or exceeding the minimum observed size for female reproduction appears to have been essentially species-specific, perhaps related to some unknown environmental factor(s). The proportion of fertile colonies does not correlate with any of the other characteristics measured or scored in this study (Table 4). This lack of correlation is most notable for brood chamber shape, mean colony size, and proportion of mckinney and taylor: bryozoan life histories 551 the colony surface area devoted to brooding, because one would expect that there would be a discernible pattern relating each of these with the proportion of fertile colonies. The observed lack of pattern may be due more to collector bias than to reality if there was a difference in preference for fertile versus non-fertile colonies among the various collectors who provided the bulk of the material for each of the species (other than the three species for which it is known that all specimens were kept when encountered). In particular, one would predict a general pattern of higher reproductive effort for smaller- growing, more ‘opportunistic’ species than for larger-growing species in which colonies might have been longer-lived. Neither of the two skeletal measures of reproductive effort - proportion of fertile colonies, and proportion of the colony surface area devoted to brooding - correlates significantly with any of the measures of colony size (Table 4). There is the possibility that with a larger sample of species, the significance (p < 0T0) of the negative correlation determined in this study between proportion of the colony surface area devoted to brooding and colony size (maximum diameter, area) might improve to acceptable significance. Life histories of small clones: constrained , plastic or hot hi There are several facets of life history theory pertinent to small clones. The theories can be broadly separated into two groups: those that seek to explain the organism’s response to prevalent environmental conditions, and those that seek to explain the effect on life history of the organism’s clonal growth. The concept of ‘spot colonies’ was introduced for colonies of encrusting ‘species settling in small spatially predictable refuges and growing to small, early maturing colonies of determinate or semi- determinate size’ (Bishop 1989, p. 214). Spot colonies are essentially equidimensional. Their attributes are well demonstrated by two cheilostome species: Cribrilina puncturata that grew preferentially on the concave surfaces of bivalve shells in the Plio-Pleistocene Red Crag of England, and which became reproductive when the colony had grown to about ten zooids (Bishop 1994); and Celleporella hyalina, most commonly growing on the alga Laminaria saccharina , which becomes reproductive when the colony reaches about 57 zooids (Cancino and Hughes 1987). Reproduction at small size and overall small total body size are characteristic of ‘/--selection’, which forms an end member of the highly debated, and currently less favoured than previously, conceptual system of life history strategies varying from long-lived, successful competitors in stable environments (A'-selection) to early-reproducing, poor competitors in uncertain environments (/•-selection) (see MacArthur and Wilson 1967;Pianka 1970; Stearns 1992). In the system of /•- versus .^-selection, small clones on substrata such as shell debris that may be intermittently disturbed are expected to have other attributes of /--selection, including rapid development, early reproduction, and semelparity, but during their single period of reproduction the capacity for production of a large number of offspring compared with more long-lived relatives that produce fewer offspring at any given period of reproduction. Another way to conceive of life history attributes of clonal animals is to examine the trade-off between early and late sexual maturity in the context of clonal growth. Timing of sexual maturity in clonal animals has the potential for extreme flexibility because the germ line is not sequestered during early growth of the organism but can be developed from one or more groups of somatic cells as the clone grows (reviewed in Buss 1987). While the modules (e.g. zooids in bryozoans) within clones eventually senesce, the clones themselves are theoretically immortal (Jackson 1985; Orive 1995), and often reach great ages (Cook 1983; Jackson 1983). For many clonal species, the larger an individual clone becomes, the higher its chance of surviving (Jackson 1985; Jackson and Coates 1986; Harvell and Grosberg 1988). Therefore, if fitness can be increased by delaying reproduction, selection may favour indefinite delay. Sexual reproduction consumes energy that in clones could otherwise go into clonal growth, i.e. asexual increase in the number of modules. Therefore, sexual reproduction at any stage in the growth of a clone should temporarily decrease or arrest growth, so that at some later stage the clone 552 PALAEONTOLOGY, VOLUME 40 will be smaller than it would have been if sexual reproduction had not occurred. An intriguing exception was noted for the arborescent cheilostome bryozoan Bugula neritina, with an unexplained increase in growth rate occurring simultaneously with reproduction (Keough 1989). Where reproductive output scales linearly or geometrically with size, as it does in many clonal animals (e.g. Gordon 1970; Hayward and Ryland 1975; Dyrynda and King 1982; Bishop 1994), it would benefit the clone to reach a maximum possible size before reproducing sexually. Such delay can be indexed to a specific size (or age) of the clone at which point reproduction occurs; the extent of the delay in reproduction would be a trade-off between probability of survival to that size-stage, versus the rate of increase in reproductive potential with increased size. Mortality of non-senescent clones is due to extrinsic, environmental factors. Therefore, for potentially immortal clones of a species living in stable environments, reproduction may be delayed until a very large size has been reached, at which point reproduction occurs for some proportion of the clones (Winston and Jackson 1984; Jackson and Wertheimer 1985). Alternatively, clones may defer sexual reproduction until some external cue is received that stimulates reproduction. The most obvious cues would be those that signal imminent total or partial mortality of the clone, such as proximity of predators or of powerful competitors, crowding by conspecifics, some temperature threshold, or physical disturbance. Such flexibility in timing of reproduction, while rare in aclonal animals, should be favoured in clonal animals ‘in which (1) the size-dependent fecundity benefits of postponing reproduction can increase without intrinsic limit and (2) the cumulative risk of reproductive failure through genet mortality increases very slowly with genet size’ (Harvell and Grosberg 1988, p. 1859). Consequently, in many species of clonal animals, once the minimum size (or age) threshold is passed, reproduction should be initiated over a broad range of sizes and should correlate neither with size nor age but instead should reflect a ‘complex interaction between intrinsic factors such as size, age, and physiological condition, as well as extrinsic factors such as density, food availability, physical disturbance, and predation’ (Harvell and Grosberg 1988, p. 1862). Individual life histories and patterns of life histories of Mesozoic encrusting cyclostomes described above are in general more consistent with hypotheses of flexibility in life history attributes, especially in flexibility in timing of reproduction, than with the concepts of r- and A'-selection. Among the attributes of r-selection seen in the Mesozoic encrusting cyclostomes are (1) small body size, although there is a relatively broad range in mean size from species-to-species and even in the smallest species some colonies extended over an order of magnitude larger than the average area; (2) semelparity; and (3) a potentially large number of sexual offspring. Although there are few gonozooids per colony in each of the species studied, polyembryony is characteristic of cyclostomes, with up to at least 11 primary embryos per brood chamber (Harmer 1896) and the capacity for a primary embryo to divide into over 100 secondary embryos (Borg 1926, p. 425). In addition, a minority of the species (e.g. Actinopora disticha) tended to reproduce at a small size rather than across a broad range of sizes. The prevalence of semelparity in most of the Mesozoic encrusting species was not associated with reproduction at some specific small colony size. Most colonies that reached the size at which reproduction could occur within their species failed to produce brood chambers. Instead, the first (and usually only) reproductive period was initiated across a range of colony sizes. In some instances, crowding with conspecifics or contact with another species or some obstruction can be seen to have been accompanied by production of brood chambers, and colonies typically grew only a short distance beyond the point at which brood chambers were completed. All of these observations are consistent with a high degree of flexibility in timing of reproduction, with reproduction stimulated by some extrinsic factor in the environment. Interaction between colony size and extrinsic factors appears to have triggered the onset of reproduction in Actinopora disticha , Liripora complanata and Plagioecia ? reniformis. In these species, the average sizes of fertile colonies are significantly greater than the average sizes of nonfertile colonies, indicating that as the colony increased in size (or age) some intrinsic factor interacted with extrinsic factors to increase the probability of the onset of reproduction. For other McKinney and taylor: bryozoan life histories 553 species, for which there is no statistical difference in size of fertile and non-fertile colonies (Table 4), nor a strongly left-skewed distribution of distances from the point of colony origin to ooecio- pores, increased colony size apparently did not increase the probability of extrinsic factors stimulating onset of reproduction. Production of brood chambers upon contact with other organisms is evidence that variation in size of colonies at time of reproduction is due largely to environmental factors rather than to large variations in growth rate of colonies with reproduction set to occur after a certain time has elapsed. CONCLUSIONS 1. Mesozoic encrusting cyclostomes show variable patterns of size-related survivorship; some species exhibit approximately constant mortality rates per unit size, whereas others have increasing mortality rates with size. In none of the species is there evidence for a fixed maximum size beyond which colonies did not grow. Even in species in which mortality rate increased with size, a few ‘Methusalah’ colonies avoided death and achieved a large size. 2. Several species were capable of producing subcolonies, usually originating at the growing edge of the parent colony but occasionally through frontal budding on to the upper colony surface. There may be a correlation between the ability to produce frontal subcolonies and to elevate the colony margin; both are traits that prevent or retard overgrowth by competitors. Species in this study lacking subcolonies achieved larger colony sizes than those with subcolonies. 3. Colony size at the onset of female sexual reproduction (as indicated by the development of brood chambers) was relatively constant in some species (e.g. Actinopora disticha) but very variable in others (e.g. Reptomultisparsa hybensis). It seems likely that environmental stimuli, such as contact with a neighbouring encruster, triggered sexual reproduction and were responsible for the observed flexibility in the timing of reproduction. 4. Colony growth typically continued for only a short time after female sexual reproduction, as indicated by the close proximity of brood chambers to colony growing edges. However, species characterized by large colony size or with longitudinally elongate brood chambers typically grew further than those with small colony size or transversely elongate brood chambers. Most colonies appear to have been semelparous with one female reproductive period, although some colonies of a few species were iteroparous with two periods. 5. Between-species comparisons of reproductive effort show no correlation between colony size and the proportion of colonies having brood chambers, or between colony size and the proportion of the colony surface occupied by brood chambers. Species with relatively large zooids reproduced at larger colony sizes and survived for longer after reproducing. 6. Mesozoic multiserial encrusting cyclostomes exhibit considerable flexibility in timing of reproduction and other life history attributes. Although having some attributes of r-selection, they do not conform to the rigid concepts of either /•- or A-selection. This supports Winston's ( 1 98 1 A) view that life history patterns in clonal organisms cannot be explained in any single or simple way. Acknowledgements. FKM is grateful to the US-UK Fulbright Commission, and to Wolfson College and the Department of Earth Sciences, University of Cambridge for funding and facilities. We thank the British Council for providing financial support for PDT to visit Slovakia, and Josef Michalik (Bratislava) for field guidance in the Triassic of the Carpathians. REFERENCES bertling, m. 1994. Okologie und Taxonomie koralleninkrustierender Bryozoen des norddeutschen Malm. Palaontologische Zeitschrift , 68, 419-435. bishop, j. d. d. 1989. Colony form and the exploitation of spatial refuges by encrusting Bryozoa. Biological Reviews , 64, 197-218. 1994. Survival and reproductive output in relation to substrate type in a bryozoan encrusting disarticulated bivalve shells. 23-28. In hayward, p. j., ryland, j. s. and taylor, p. d. (eds). Biology and palaeobiology of bryozoans. Olsen & Olsen, Fredensborg, 240 pp. 554 PALAEONTOLOGY, VOLUME 40 borg, F. 1926. Studies on Recent cyclostomatous Bryozoa. Zoologiska Bidrag fran Uppsala , 10, 181-507. brood, k. 1972. Cyclostomatous Bryozoa from the Upper Cretaceous and Danian in Scandanavia. Stockholm Contributions in Geology , 26, 1-464. 1981. Two new Maastrichtian species of Serpentipora (Bryozoa). Palaontologische Zeitschrift , 55, 1365-139. buss, l. w. 1981. Mechanisms of competition between Onychocella alula (Hastings) and Antropora tincta (Hastings) on an eastern Pacific rocky shoreline. 39-49. In larwood, g. p. and nielsen, c. (eds). Recent and fossil Bryozoa. Olsen & Olsen, Fredensborg, 334 pp. 1987. The evolution of individuality. Princeton University Press, Princeton, 201 pp. cancino, j. m. and hughes, r. n. 1987. The effect of water flow on growth and reproduction of Cellepore/la hyalina (L.) (Bryozoa: Cheilostonrata). Journal of Experimental Marine Biology and Ecology , 1 12, 109-130. — Castaneda, b. and orellana, M. c. 1991. Reproductive strategies in bryozoans: experimental test of the effects of conspecific neighbours. Bulletin de la Societe des Sciences Naturelles de TOuest de la France, Memoire Hors Serie, 1, 81-88. chadwick-furman, n. E. and weissman, i. l. 1995. Life histories and senescence of Botryllus schlosseri (Chordata, Ascidiacea) in Monterey Bay. Biological Bulletin , 189, 36-41. cook, R. G. 1983. Clonal plant populations. American Scientist , 71, 244-253. craig, G. c. and oertel, g. 1966. Deterministic models of living and fossil populations of animals. Quarterly Journal of the Geological Society, London, 122, 315-355. Dudley, j. e. 1973. Observations on the reproduction, early larval development, and colony astogeny of Conopeum tenuissimum (Canu). Chesapeake Science, 14, 270-278. dyrynda, p. e. j. and king, p. e. 1982. Sexual reproduction in Epistomia bursaria (Bryozoa: Cheilostomata), an endozooidal brooder without polypide recycling. Journal of Zoology, London, 198, 337-352. 1983. Gametogenesis in placental and non-placental ovicellate cheilostome Bryozoa. Journal of Zoology, London, 200. 471^192. and ryland, j. s 1982. Reproductive strategies and life histories in the cheilostome bryozoans Chartella papyracea and Bugula flabellata. Marine Biology, 71, 241—256. eggleston, d. 1972. Patterns of reproduction in marine Ectoprocta of the Isle of Man. Journal of Natural History, 6, 31-38. flessa, k. w., cutler, a. H. and meldahl, K. h. 1993. Time and taphonomy: quantitative estimates of time- averaging and stratigraphic disorder in a shallow marine habitat. Paleobiology, 19, 266-286. Gordon, d. p. 1970. Reproductive ecology of some northern New Zealand Bryozoa. Cahiers de Biologie Marine, 11, 307-323. GREGORY, J. w. 1899. Catalogue of the fossil Bryozoa in the Department of Geology , British Museum ( Natural History). The Cretaceous Bryozoa. Volume 1. British Museum (Natural History), London, 457 pp., 17 pis. hagenow, F. von 1851 Die Bryozoen der Maastrichter Kreidebildung. Fischer, Cassel, xv+111 pp., 12 pis. hakansson, E. 1976. Population structure of colonial organisms. A palaeoecological study of some free-living bryozoans. Documents des Laboratoires de Geologie de la Faculte des Sciences de Lyon, Hors Serie, 3, 385-399. harmer, s. F. 1896. On the development of Lichenopora verrucaria, Fabr. Quarterly Journal of the Microscopical Society, 39, 71-144, pis 7-10. harvell, c. d. and grosberg, R. K. 1988. The timing of sexual maturity in clonal animals. Ecology, 69, 1855-1864. hayward, p. j. and ryland, j. s. 1975. Growth, reproduction and larval dispersal in Alcyonidium hirsutum (Fleming) and some other bryozoans. Pubblcazioni della Stazione Zoologica di Napoli, 39 (Suppl.), 226-241. herrera, a., jackson, j. b. c., hughes, d. J., jara, J. and ramos, h. 1996. Life-history variation in three coexisting cheilostome species of the bryozoan genus Stylopoma in Panama. Marine Biology, 126, 461-469. holder, h. 1972. Endo- and Epizoen von Belemniten-Rostren ( Megateuthis ) im nordwestdeutschen Bajocium (Mittlerer Jura). Palaontologische Zeitschrift, 46, 199-220. hughes, t. p. and jackson, j. b. c. 1980. Do corals he about their age? Some demographic consequences of partial mortality, fission, and fusion. Science, 209, 713-715. hunter, e. and hughes, r. n. 1993. Effects of diet on life-history parameters of the marine bryozoan, Celleporella hyalina (L.). Journal of Experimental Marine Biology and Ecology, 167, 163-177. jackson, j. b. c. 1983. Biological determinants of present and past sessile animal distributions. 39-120. In tevesz, m. j. s. and McCall, p. l. (eds). Biotic interactions in Recent and fossil benthic communities. Plenum Publishing Corporation, New York, 837 pp. 1985. Distribution and ecology of clonal and aclonal benthic invertebrates. 297-355. In jackson, j. b. c., McKINNEY AND TAYLOR: BRYOZOAN LIFE HISTORIES 555 buss, L. w. and COOK, r. e. (eds). Population biology and evolution of clonal organisms. Yale University Press, New Haven, 530 pp. — ■ and coates, a. g. 1986. Life cycles and evolution of clonal (modular) animals. Philosophical Transactions of the Royal Society of London, Series B , 313, 7-22. — and Wertheimer, s. p. 1985. Patterns of reproduction in five common species of Jamaican reef-associated bryozoans. 161-168. In nielsen, c. and larwood, g. p. (eds). Bryozoa: Ordovician to Recent. Olsen & Olsen, Fredensborg, 364 pp. and winston, j. E. 1981. Modular growth and longevity in bryozoans. 121-126. In larwood, g. p. and nielsen, c. (eds). Recent and fossil Bryozoa. Olsen & Olsen, Fredensborg, 334 pp. Johnson, G. a. l. and nudds, j. r. 1975. Carboniferous coral geochronometers. 27-42. In rosenberg, g. d. and runcorn, s. K. (eds). Growth rhythms and the history of the Earth's rotation. Wiley, London, 559 pp. karande, a. a. and udhayakumar, m. 1992. Consequences of crowding on life-histories of cheilostome bryozoans in Bombay waters. Indian Journal of Marine Sciences, 21, 133-136. keough, m. j. 1989. Variation of growth rate and reproduction of the bryozoan Bugula neritina. Biological Bulletin , 177, 277-286. lasker. h. r. 1983. Vegetative reproduction in the octocoral Briareum asbestinum (Pallas). Journal of Experimental Marine Biology and Ecology, 72, 157-169. levitan, D. R. and petersen, c. 1995. Sperm limitation in the sea. Trends in Ecology and Evolution, 10, 228-23 1 Macarthur, r. h. and wilson, e. o. 1967. Theory of island biogeography. Princeton University Press, Princeton, 203 pp. martill, D. M. and HUDSON, J. D. 1991 . Fossils of the Oxford Clay. Palaeontological Association Field Guide to Fossils Number 4. Palaeontological Association, London, 286 pp. mawatari, s. f. 1975. The life history of Membranipora serrilamella Osburn (Bryozoa, Cheilostomata). Bulletin of the Liberal Arts & Science Course , School of Medicine Nihon University, 3, 19-57, pis 1-5. mayoral, E. and sequeiros, l. 1991. Significado paleoecologico de algunos epizoos y ‘borers’ del Jurasico Inferior y medio de Belchite (Zaragoza, Cordillera Iberica). Caudernos de Geologia, 10 [for 1979], 121-135. McKinney, F. k. 1983. Asexual colony multiplication by fragmentation: an important mode of genet longevity in the Carboniferous bryozoan Archimedes. Paleobiology, 4, 35-43. - 1992. Competitive interactions between related clades: evolutionary implications of overgrowth interactions between encrusting cyclostome and cheilostome bryozoans. Marine Biology, 114, 645-652. - 1993. A faster-paced world?: contrasts in biovolume and life-process rates in cyclostome (Class Stenolaemata) and cheilostome (Class Gymnolaemata) bryozoans. Paleobiology, 19, 335-351. — galloway, G. m. and McKinney, m. j. 1996. Colony shapes and sizes: some life-history attributes of encrusting cyclostomate bryozoans (Eocene, North Carolina). 179-185. In Gordon, d. p., smith, a. m. and grant-mackie, j. a. (eds). Bryozoans in space and time. National Institute of Water & Atmospheric Research, Wellington, 442 pp. and jackson, j. b. c. 1989. Bryozoan evolution. Unwin Hyman, Boston, 238 pp. ORBIGNY, A. d' 1850. Prodrome de paleontologie stratigraphique universelle des animaux Mollusques et rayonnes. Tome 1. Masson, Paris, 394 pp. 1851-1854. Paleontologie frangaise. Terrain cretace. Tome cinquieme, Bryozoaires. Texte et Atlas. Masson, Paris, pp. 1-188 (1851), 189-472 (1852), 473-984 (1853), 985-1192 (1854); pis 600-683 (1851), 684-761 (1852), 762-800 (1853). orive, m. e. 1995. Senescence in organisms with clonal reproduction and complex life histories. American Naturalist, 145, 90-108. palmer, t. j. and fursich, f. t. 1974. The ecology of a Middle Jurassic hardground and crevice fauna. Palaeontology, 17. 507-524. 1981. Ecology of sponge reefs from the Upper Bathonian of Normandy. Palaeontology, 24, 1-23. — and wilson, m. a. 1990. Growth of ferruginous oncoliths in the Bajocian (Middle Jurassic) of Europe. Terra Nova, 2, 142-147. pianka, E. r. 1970. On r- and A-Selection. American Naturalist, 104. 592-597. poluzzi, a. and sartori, r. 1975. Report on the carbonate mineralogy of Bryozoa. Documents des Laboratoires de Geologie de la Faculte des Sciences de Lyon, Hors Serie, 3, 193-210. prantl, f. 1938. Erster Fund von Bryozoen in dem karpathischen Rhat. Zentralblatt fiir Geologie und Paldontologie, Abhandlungen B, 7, 262-264. roemer, f. a. 1840. Die Versteinerungen des Norddeutschen Kreidegebirges. Erste Lieferung. Hahn'schen Hofbuchhanglung, Hannover, iv + 48 pp., 7 pis. ryland, j. s. 1963. Systematic and biological studies on Polyzoa (Bryozoa) from western Norway. Sarsia, 14, 1-59. 556 PALAEONTOLOGY, VOLUME 40 1970. Bryozoans. Hutchinson University Press, London, 175 pp. 1996. Polyembryony ‘paradox’: the case of cyclostomate Bryozoa. Trends in Ecology and Evolution , 1 1, 26. sackville Hamilton, n. r., schmid, b. and harper, j. l. 1987. Life-history concepts and the population biology of clonal organisms. Proceedings of the Royal Society of London, Series B , 232, 35-57. schafer, p. 1991. Brutkammern der Stenolaemata (Bryozoa): Konstruktionsmorphologie und phylogenetische Bedeutung. Courier Forschungsinstitut Senckenberg , 136, 1-263. silen, l. 1966. On the fertilization problem in the gymnolaematous Bryozoa. Ophelia, 3, 1 13-140. 1972. Fertilization in the Bryozoa. Ophelia, 10, 27-34. 1977. Polymorphism. 183-231. In woollacott, r. m. and zimmer, r. l. (eds). Biology of bryozoans. Academic Press, New York, 566 pp. stebbing, a. r. d. 1973. Observations on colony overgrowth and spatial competition. 173-183. In larwood, g. p. (ed.). Living and fossil Bryozoa. Academic Press, London, 634 pp. Stearns, s. c. 1992. The evolution of life histories. Oxford University Press. Oxford, 249 pp. strom, r. 1977. Brooding patterns of bryozoans. 23-55. In woollacott, r. m. and zimmer, r. l. (eds). Biology of bryozoans. Academic Press, New York, 566 pp. taylor, p. d. 1979. Functional significance of contrasting colony form in two Mesozoic encrusting bryozoans. Palaeogeography, Palaeoclimatology, Palaeoecology, 26, 151-158. and michalik, j. 1991 . Cyclostome bryozoans from the late Triassic (Rhaetian) of the West Carpathians, Czechoslovakia. Neues Jahrbuch fur Geologie und Paldontologie, 182, 285-302. vail, l. l. and wass, r. e. 1 98 In. Experimental studies on the settlement and growth of Bryozoa in the natural environment. Australian Journal of Marine and Freshwater Research, 32, 639-656. 1981b. Experimental studies on the settlement and growth of Bryozoa in the natural environment. Accessory Publication, Tables 2-5. University of Sydney, Department of Geology, 4 pp. [unnumbered], vermeij, G. j. 1987. Evolution and escalation. Princeton University Press, Princeton, 527 pp. voigt, E. 1962. Verkhnemelovye mshanki evropeyskoy chasti SSSR i nekotorykh soprede! nykh oblastey. Izdatel’stvo Moskovskogo Universiteta, Moscow, 125 pp. [In Russian], Walter, b. 1989. Les Diastoporidae bereniciformes neocomiens du Jura Franco-Suisse. Palaeontographica, Abteilung A, 207, 107-145. Walters, l. j. and wethey, d. s. 1986. Surface topography influences competitive hierarchies on marine hard substrata: a field experiment. Biological Bulletin. 170, 441 — 449. wilson, m. a. 1986. Coelobites and spatial refuges in a Lower Cretaceous cobble-dwelling hardground fauna. Palaeontology, 29, 691-703. winston, j. e. 1977. Feeding in marine bryozoans. 34-57. In woollacott, r. m. and zimmer, r. l. (eds). Biology of bryozoans. Academic Press, New York, 566 pp. 198 la. Feeding behavior of modern bryozoans. University of Tennessee Department of Geological Sciences Studies in Geology, 5, 1-21. 1981b. Life histories of colonial invertebrates. Paleobiology, 7, 151-153. 1985. Life history studies of Disporella and Drepanophora in Jamaica. 350. In nielsen, c. and larwood, G. p. (eds). Bryozoa : Ordovician to Recent. Olsen & Olsen, Fredensborg, 364 pp. and hakansson, E. 1986. The interstitial bryozoan fauna from Capron Shoal, Florida. American Museum Novitates, 2865. 1-50. — and jackson, J. B. c. 1984. Ecology of cryptic coral reef communities. IV. Community development and life histories of encrusting cheilostome Bryozoa. Journal of Experimented Marine Biology and Ecology, 7, 1-21. FRANK K. McKINNEY Department of Geology Appalachian State University Boone, North Carolina 28608, USA PAUL D. TAYLOR Department of Palaeontology The Natural History Museum Typescript received 26 September 1995 Cromwell Road Revised typescript received 30 April 1996 London SW7 5BD, UK THE DIET OF THE EARLY TOARCI AN AMMONITE HARPOCERAS FALCIFERUM by MANFRED JAGER and RENE FRA A YE Abstract. Diagenetically compressed ammonites from the Early Toarcian Posidonienschiefer in southern Germany yield new data on the diet and ingestion regulation of ammonites. About 4 per cent, of the relatively large body chambers of adult Harpoceras falciferum macroconchs contain distinctive food remains, mostly pereiopods of small decapod crustaceans, which probably were the main prey of this ammonite species, and rarely abdomens and telsons of the same crustaceans or aptychi of small ammonites. The contents of the digestive tract are preserved in the adapical three-quarters of the body chamber as a row of ‘food balls’. The number of these food balls is variable; up to live have been counted. In most specimens it is not possible to distinguish between crop and stomach contents. About 1 per cent, of adult Harpoceras falciferum macroconchs contain bivalve debris in their body chamber. Although in some specimens this may represent the crop content of the ammonite, in the majority of specimens the debris may be interpreted as food remains not of the ammonite itself, but of another animal living in the mantle cavity area of dead ammonites. Ammonites are one of the best documented fossil groups, with an extensive literature on morphology, evolution and palaeoecology. However, data on feeding mechanisms and diet are sparse. The aim of this paper is to give new informations about the diet of an important Early Jurassic species, Harpoceras falciferum, by investigation of the contents of the body chamber. DIET AND FEEDING OF RECENT NAUTILUS Today, most cephalopods are highly skilled carnivores, with a very efficient digestive system, the rate of ingestion ranging between three and 20 hours. The majority take a wide variety of prey; mainly crustaceans, molluscs and fishes and to a lesser extent echinoderms, polychaetes, chaetognaths and siphonophores. The prey may be ingested whole, or in bite-size pieces either including the exoskeleton or the flesh exclusively (Nixon 1987, 1988). Modern Nautilus , which is similar to the ammonites in its exoskeleton and in many other respects, scavenges both exuviae and dead decapod crustaceans. Although Nautilus catches live prey when enclosed together in traps or aquariums, it has never been observed to catch live prey if this is able to escape. Different species of crustaceans are the main food source for Nautilus (Ward 1987, pp. 156-159). ‘At every locale where crop contents or faecal residues have been examined, crustacean test fragments were recorded’ (Saunders and Ward in Saunders and Landman 1987, p. 151). Ward and Wicksten (1980) identified 23 crustacean specimens (mostly Aniculus anicuius , a hermit pagurid crab, but also brachyuran and raninid crabs, a galatheid and fragments of a palinurid lobster, as well as fish bones) as crop, stomach, and intestine contents in nine specimens of Nautilus macromphalus from New Caledonia. According to Nixon (1988, p. 645), Nautilus bites crustaceans ‘into pieces of about 5 mm3, which could pass along the highly distensible oesophagus (Haven 1972) to be stored in the large crop’. Nautilus macromphalus from New Caledonia often eats fresh moults of larger crustaceans, such as lobsters (e.g. Panulirus longipes ), Ward (1987, p. 159) referred to observations by Magnier and Laboute (1978) and Ward and Wicksten (1980). Ward (1987, p. 159) assumed that there are two IPalaeontology, Vol. 40, Part 2, 1997, pp. 557-574, 8 pis) © The Palaeontological Association 558 PALAEONTOLOGY, VOLUME 40 THE DIET OF AMMONITES: PREVIOUS DATA Crop and/or stomach content (interpretation Stratigraphy and by the original locality Ammonite species authors) Author Upper Jurassic (Lower Tithonian), lithographic limestone, Solnhofen area, Bavaria, Germany Neochetoceras steraspis (Oppel) Same specimen, plus a second, similar one Numerous small, broken aptychi, interpreted as brood Ditto, interpreted as stomach content Michael (1894) Lehmann (1976, p. 129) Upper Jurassic (Upper Kimmeridgian), lithographic limestone, Nusplingen, Baden- Wiirttemberg, Germany Physodoceras sp. Same specimen Stomach content : ossicles of the crinoid Saccocoma sp. Crop content : echinoid spines Lehmann (1976, p. 129) Riegraf et al. (1984, p. 58) Lower Jurassic Hildoceras ( Hildaites ) Stomach content : Lehmann and Weitschat (Lower Toarcian), Posidonienschiefer, Haverlahwiese, Lower Saxony, Germany levisoni (Simpson) jaw apparatus of a small ammonite (1973) Posidonienschiefer, Phylloceras Crop content : debris of Riegraf et al. (1984, Unterer Schiefer, Dotternhausen, Baden- Wiirttemberg, Germany heterophyllum (J. Sowerby) Pseudomytiloides dubius (J. de C. Sowerby), shells, and doubtful echinoderm remains pp. 53-54, 57) Posidonienschiefer, Harpoceras falciferum Crop content: doubtful Riegraf et al. (1984, same locality (J. Sowerby) debris of Pseudo- mytiloides dubius p. 56) Posidonienschiefer, Hildoceras ( Hildaites ) Crop content: small Riegraf et al. (1984, Schierferklotz layer, Ohmden, Baden- Wiirttemberg, Germany serpentinum (Reinecke) aptychus pp. 54-56) Posidonienschiefer, Hildoceras (Hildaites) Stomach content: calcitic Riegraf et al. (1984, Unterer Schiefer, Dotternhausen, Baden- Wiirttemberg, Germany levisoni (Simpson) A second specimen debris resembling echinoderm stereom Stomach content: aragonitic shell debris p. 57, 189) Riegraf et al. (1984, p. 57) [continued on p. 560 EXPLANATION OF PLATE 1 Figs 1-4. Distal parts of pereiopods with chelae of Coleial sp. in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. I , many chelae; W 60; x 1-6. 2, single chela; W 50; x 2-7. 3, many chelae; W 68; x 2-4. 4, single chela; W 63; x 2-8. PLATE 1 JAGER and FRAAYE, body chamber contents of Harpoceras 560 PALAEONTOLOGY, VOLUME 40 THE DIET OF AMMONITES: PREVIOUS DATA (Com.) Crop and/or stomach content (interpretation Stratigraphy and by the original locality Ammonite species authors) Author Lower Jurassic, Aniioceras sp. Sinemurium, ?Yorkshire, England Stomach content: Lehmann (1972) foraminifers and ostracods Triassic, Spitzbergen Sv alb ar dicer as spitzbergense (Frebold) Stomach content : Lehmann ( 1985, p. 102; many fragments of 1990, p. 184) ostracods main reasons why Nautilus ingests moults : to extract the protein from the extensive integuments which line the internal parts of the moults and/or the calcium for shell building. Ingestion of lobster exuviae follows a specific pattern in which Nautilus consumes the exoskeleton beginning at the posteriormost part of the abdomen and continuing anteriorly (Ward and Wicksten 1980; Tshudy et al. 1989). In laboratory feeding experiments pereiopods are often seized first, with biting concentrated on the tips of the pleopods and the softer underside of the abdomen. When Nautilus reaches the telson, it eats the uropods and subsequently the abdomen segment by segment. The heavily calcified cephalothorax is left behind (Ward 1987, p. 159). On the other hand. Ward and Wicksten (1980) reported finding well-worn pieces of crustacean carapace material in the crop, stomach and caecum of TV. pompilius from Fiji and TV. macromphalus from New Caledonia. An adult Nautilus feeds on a large lobster exuvia for as long as two hours (aquarium experiment. Ward and Wicksten 1980). Saunders and Ward (in Saunders and Landman 1987, p. 151) reported that the dissected crops of specimens of TV. pompilius from Lae, Papua New Guinea, contained many fresh fragments of deep-water regular echinoids including both test fragments and viscera. Other specimens of TV. pompilius , dissected in Manus, contained coleoid beaks, and occasional Nautilus tentacle fragments. According to Ward (1987, pp. 159-160), Nautilus appears to be a windfall feeder. Ward et al. ( 1977) showed that a 500 g Nautilus can store up to 100 g of bait material in its crop. K. Mangold (pers. comm.) has conducted experiments on individuals of TV. macromphalus showing that food material can be stored for up to two weeks in the crop before digestion. Tanabe et al. (1980) reported that the crop is capable of enormous enlargement, as it may measure as much as 80 mm long by 50 mm in diameter when filled with food. An excellent summary was provided by Lehmann (1976); Lehmann (1990) did not refer to additional specimens, but (p. 185) wondered why no crustaceans (except for ostracods) were found in the crop or stomach contents of ammonites, especially as he thought that crustaceans were one of the main types of prey of ammonites. Bandel (pers. comm., see Riegraf et al. 1984, p. 59) supposed that the food remains found inside ammonites may have come out of the stomach of fishes, which probably had been eaten by the ammonites. EXPLANATION OF PLATE 2 Figs 1-5. Complete pereiopods with chelae of Coleiat sp. in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1-4. W-shaped pairs of pereiopods. 1, W 9; x 1-8. 2, W 20; x 1-5. 3, W 1 ; x 0-5. Arrow indicates section enlarged in fig. 4. 4, x2-2. 5, V-shaped pereiopod; W 5; x 3-0. PLATE 2 JAGER and FRAAYE, body chamber contents of Harpoceras 562 PALAEONTOLOGY, VOLUME 40 DIET OF AMMONITES FROM THE LOWER TO A RCI AN OF DOTTE R N HAUSEN In the Early Toarcian Posidonienschiefer of south-west Germany compressed body chambers occasionally preserve contents of different kinds. A single large crustacean from Dotternhausen has recently been interpreted by Fraaye and Jager (1995) to have occupied an empty body chamber. Dark homogeneous spots in ammonites from Dotternhausen and other localities were figured and described as crop and stomach contents by Riegraf et al. (1984, pp. 53-59). Such brown or (less commonly) black spots are frequently found and are present also in some of the specimens described below. However, as they are often indistinct, they are not further considered here. Ammonites are present at every level of the 9-10 m thick Posidonienschiefer facies in Dotternhausen, being common at most levels. Within a section a few decimetres thick, in the lowermost part of the commune subzone, above the ‘Inoceramenbank’, the intense red-brown colour of the periostracum is most attractive to the collector, and the recovery of complete specimens is relatively easy, as the shale often splits along bedding planes. In this part of the section collecting of ammonites has been much more intensive than in other parts of the sequence. Ammonite taxa collected from this layer are (in degree of abundance): Dactylioceras commune , Harpoceras falciferum (although being the index of the falciferum subzone, this species ranges into the commune subzone, the diameter of adult macroconchs here in general being 200-300 mm), Hildoceras ex gr. douillei / sublevisoni, Pseudolioceras lythense , Phylloceras heterophyllum , Lytoceras sp. and Phymatoceras cf. escheri. From this part of the Dotternhausen section, 72 specimens with determinable body chamber contents are described and discussed, all being adult macroconchs of Harpoceras falciferum. Remains of a small eryonid crustacean, possibly Co/eici sp., were found in 62 ammonite specimens (Pis l^t; PI. 7, fig. 1 ; PI. 8). Some of these have been previously reported by Jager (1991, pp. 33-34, erroneously interpreted as inhabitants; 1993, p. 66, fig. 49). Two of the Coleia ?-bearing specimens and nine more specimens contain concentrations of bivalve debris (Pis 5-6). A single specimen preserves several small aptychi, but neither crustacean remains nor bivalve debris, in its body chamber (PI. 7, figs 2-4), and some of the specimens with crustaceans contain small aptychi as well (PI. 7, fig. 1 ; PI. 8). All fossils, unless stated otherwise, are housed in the collection of Rohrbach Zement in Dotternhausen. Specimens with crustacean remains (Coleia? sp.) In the part of the section described above, roughly 4 per cent, of adult Harpoceras falciferum macroconchs (estimated during excavation in the quarry: three out of 77 specimens) or even more in the lowermost part of the section (estimated from a later excavation without counting) preserve crustacean remains in their body chamber. In other co-occurring ammonite genera, crustacean remains are extremely rare; the few specimens will be described elsewhere. No crustaceans outside ammonite shells have been found. However, the ‘inhabitant’ Palaeastacusl described by Fraaye and Jager ( 1995) comes from this level, and bite marks in some ammonite shells from the same layer (e.g. Jager 1991, fig. 3) provide indirect evidence for the presence of large crustaceans. In specimens of Harpoceras falciferum from lower stratigraphical levels no crustaceans have been found. EXPLANATION OF PLATE 3 Figs 1-4. Food balls including fragments of pereiopods, some of them with chelae, of Coleia ? sp. in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. The broad chelae indicated by arrows (figs 1-2) may belong to a different genus. 1, W 4; x 2-2. 2, W 48; x 2-9. 3, W 3; x2-4. 4, W 18; x 3 0. PLATE 3 JAGER and FRAAYE, body chamber contents of Hcirpoceras 564 PALAEONTOLOGY, VOLUME 40 These observations allow the following conclusions to be drawn : 1 . Harpoceras falciferum, like modern Nautilus , frequently fed upon crustaceans (dead or exuviae?), whereas representatives of other ammonite genera normally did not. This underscores suppositions of a number of authors that living conditions of various genera differed. 2. Coleial must have lived in separate geographical regions (probably nearer to the coast, perhaps in Franconia, where crustaceans are more often found than in Swabia), and Harpoceras falciferum must have eaten them there and subsequently returned over some tens or even more than a hundred kilometres to the basin area around Dotternhausen. 3. As specimens of Harpoceras falciferum from lower strata do not preserve crustaceans inside their body chamber, either the preferred kind of food or the conditions of preservation of food remains must have changed. In 54 of the 62 specimens with crustaceans, the body chamber is complete, or the region where the aptychi should be expected at least is preserved. Ten of these 54 specimens (= 18-5 per cent.) preserve aptychi. This is nearly the double percentage than in the total number of adult Harpoceras falciferum macroconchs with or without crustaceans from the same strata : only eight out of 77 Harpoceras specimens (= 10-4 per cent.) preserve their aptychi in the body chamber. Thus, in specimens without crustaceans, the contents of the body chamber has decayed to a greater extent than in specimens with crustaceans. This means that originally more than the mere 4 per cent, of Harpoceras had crustaceans in their digestive tract, and that many lost their digestive tract together with their aptychi during decay of the soft parts. All crustaceans in Harpoceras falciferum body chambers (except for the Palaeastacusl described by Fraaye and Jager 1995) are of comparable size, most specimens being similar to Proeryon , but distinctly smaller. Thus assignment to Coleia seems probable, although diagnostic features cannot be made out. However, it cannot be ruled out that they represent juveniles of Proeryon which lived near-shore. Most Coleial specimens are preserved as indistinct remains of pereiopods (PI. 3, figs 1-4), and of distal parts of pereiopods with chelae (PI. 1, figs 1-4; PI. 4, fig. 2), but sometimes V-shaped complete pereiopods including chelae (PI. 2, fig. 5) or even W-shaped pairs of pereiopods with chelae (PI. 2, figs 1-4) are seen. Few chelae are broader than the others and may represent a different genus (PI. 3, figs 1-2). Remains of the abdomen and telson (PI. 4, figs 3-6) are much rarer than pereiopods. Not a single remain of the cephalothorax has been identified. This pattern is in full accordance with the pattern of selective consumption in Recent Nautilus described above. However, the complete pereiopods or pairs of pereiopods are much larger than the 5 mm3 bites that Recent Nautilus manages. 'The hard, calcified cutting edges found on the jaws of Nautilus are used to break up hard crustacean carapace material; the very different jaw edge morphologies of the Jurassic ammonites so far studied suggests differences in food source’ (Ward 1987, p. 248). Although this last conclusion is questioniable, Harpoceras falciferum at least was unable to bite the crustaceans into such small pieces. Some questions, however, remain unresolved. For example, it is not quite clear yet if the aptychi were used for biting or as a 'shovel’. Dagys et al. (1989, pp. 49-50) suggested differences in the function of anaptychi and aptychi: ‘The main distinguishing feature of the anaptychus-type lower jaw is absence of calcareous coverings and presence of a more or less marked pit in the rostral part EXPLANATION OF PLATE 4 Figs 1-6. Remains of Coleial sp. in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1, large masses of pereiopods; W 40; x 1-3. 2, fragments of pereiopods with chelae; W 27; x 2-6. 3-6, fragments of abdomen and telson. 3, W 17; x 4-6. 4, W 18; x 3-0. 5, negative impression of specimen in fig. 4; W 18; x 3-5. 6, W 62; x 4-6. PLATE 4 JAGER and FRAAYE, body chamber contents of Harpoceras 566 PALAEONTOLOGY, VOLUME 40 of the inner lamella. It is likely that the upper jaw was overlapped by the lower jaw, as in recent cephalopods, and that the pit represents the place of insertion of the rostrum of the upper jaw. If this reconstruction is correct, it may be supposed that the main action of this type of jaw apparatus was crushing. In this case the older ammonoids may have preferred rather coarse food and animals with fairly hard shells. This specialization was absent in the aptychus-type of jaw apparatus, which is characterized by the presence of calcareous plates covering the flanks of the lower jaw. Their front edges may have supported a cutting function of the jaw, at least more so than in the anaptychus type. But the earlier suggestion of a shovel-like function (Lehmann 1972, 1975, 1981), without too much crushing or cutting, still has its merits. Morton and Nixon (1987) suggested that the shovel- like lower jaws may have expelled water while retaining captured small prey.’ If the aptychi acted as a ‘shovel' only, how did Harpoceras manage to separate pereiopods from the cephalothorax? It cannot be ruled out totally that the crustaceans were consumed not by the ammonite, but by an unknown post-mortem occupant of the empty ammonite shell. There are, however, arguments contradicting this alternative hypothesis. 1. One should expect that a hypothetical occupant should throw the aptychi out of the shell by its life activities rather than keep them inside. The fact is, however, that the aptychi are still present significantly more often in ammonites with crustaceans than in ammonites without crustaceans. 2. The fact that in Dotternhausen hitherto Coleial has not been found in the slabs outside the ammonites, and thus presumably did not have its natural habitat at this locality, raises the question of how the crustaceans were transported from their habitat to Dotternhausen, if not by an actively swimming ammonite (if the ammonite was already dead and drifted passively only, one should expect the soft body including the aptychi and the crustaceans to decay quickly and to have fallen out of the shell.) It seems improbable that an unknown animal fed upon the crustaceans in their habitat, and then swam many kilometres to Dotternhausen to hide itself in an empty ammonite shell for digestion. In this case, the crustaceans should presumably be regarded as coprolites, but don’t look like them. In the Posidonienschiefer, coprolites of carnivorous reptiles and fishes are usually preserved as three-dimensional yellow-brown phosphatic masses, whereas the ?coprolites of Palaeastacusl (Fraaye and Jager 1995) are preserved as flattened circular spots. The fact that Coleial is not found outside the ammonite shells contradicts another hypothesis: that the crustaceans were only passively drifted into the ammonite shell by currents (‘fossil trap’). There is no reason to believe that the potential of preservation of fossil crustaceans within the shell, which did not form a concretion, is greater than outside (in contrast to aragonitic gastropod shells, which are preserved only in certain calcareous concretions, but not in the shale outside). Now that quite a number of specimens has been examined, the type of crustacean remains preserved, and the notable absence of the cephalothorax, clearly counters the hypothesis that the small Coleial crustaceans either lived or moulted in empty ammonite shells (as erroneously supposed by Jager 1991). In some specimens the crustacean remains are distributed in no particular pattern in the body chamber, but more often they are concentrated in densely packed balls of varying shape and size. The normal diameter of these balls is 10-50 mm, but there are a few masses greater than 80 mm long and 50 mm wide with the crustacean remains lying in two or three different levels. Commonly, single pereiopods jut out of the compact balls. The single remains may be interpreted as being derived from disarticulated or partly disarticulated balls. When splitting up the slabs of shale, very often one part of a ball remains on one slab and the other part remains on the opposite slab, leaving a EXPLANATION OF PLATE 5 Figs 1-4. Bivalve debris in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1, W 55; the section enlarged in fig. 2 is indicated by an arrow; x 0-5. 2, x 2-5. 3, W 33; the section enlarged in fig. 4 is indicated by an arrow; x 0 5. 4, x 2-7. PLATE 5 JAGER and FRA A YE, body chamber contents of Harpoceras 568 PALAEONTOLOGY, VOLUME 40 4 1] text-fig. 1. Positions of centres of 120 food balls including Coleial sp., in 58 Harpoceras falciferum specimens, drawn in the body chamber of an idealized ammonite. complicated pattern of positive food ball fragments and negative impressions on either slab (PI. 1, figs 1, 3; PI. 3, figs 1, 3; PI. 4, fig. 1). Some of the crustacean remains are fairly distinct when viewed under low-angle light, but others may be confused with compression cracks in the ammonite shell or are hidden under epizoans or broken away. Thus it is difficult to draw the correct circumference of the balls. Nevertheless, a sketch (Text-fig. 1) shows the position of the (approximate) centre of 120 food balls (seven more balls could not be located because the anterior and posterior ends of the body chamber are not preserved.) The number of crustacean remains varies from a single pereiopod in the whole body chamber (PI. 2, fig. 5) up to five crustacean balls plus several aptychi filling a considerable part of the body chamber (PI. 7, fig. 1; PI. 8, figs 1-3). This resembles the high storage potential of Recent Nautilus mentioned above. Counting of balls is difficult, because sometimes it is impossible to decide whether a large spot represents a single large ball or results from the close proximity of two partially disarticulated balls. Twenty-five specimens preserve a single ball, in 22 specimens there are two, in six specimens three, in five specimens four, and in four specimens five. In two exceptional specimens (registered as having four and five balls, respectively), masses of crustacean remains nearly fill the entire body chamber from the last septum to the aperture, leaving only small areas free (PI. 4, fig. 1). This may even exceed the storage potential and casts doubt on the interpretation that all these masses were really consumed by the ammonite. One of these exceptional specimens does not preserve aptychi, but the other does. It may be a matter of chance that these two specimens both show bivalve encrustation. EXPLANATION OF PLATE 6 Figs 1-5. Bivalve debris in the body chamber of adult macroconchs of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1, W 73; x 1-2. 2, W 30; x 17. 3, W 31 ; the section enlarged in figs 4—5 is indicated by an arrow; xO-4. 4, x L3. 5, x 1-9. PLATE 6 JAGER and FRAAYE, body chamber contents of Harpoceras 570 PALAEONTOLOGY, VOLUME 40 If the body chamber, whose length equals nearly three-quarters of a whorl in Harpoceras falciferum , is divided into four sectors of equal length from the aperture (sector 1) to the last septum (sector 4), then seven centres of food balls are positioned in sector 1, 32 in sector 2, 47 in sector 3, and 34 in sector 4. As there is only one position of maximum abundance (sector 3), it cannot be decided beyond doubt whether the crustacean remains represent crop contents (which should be positioned in sectors 2-3) and/or stomach contents (which should be positioned in sector 4). The remains in sector 1 and possibly those in the anterior half of sector 2, too, are assumed to have been dislocated, whereas the remains in the adapical three-quarters of the body chamber are preserved probably more or less in their original position in the digestive tract. The position of the digestive tract (oesophagus median, stomach in the posterior part (sector 4) of the body chamber) was shown in Aniioceras and Hildoceras by Lehmann (1972) and Lehmann and Weitschat (1973). Specimens with bivalve debris In the Posidonienschiefer, slabs very often show remains of Pseudomytiloides dubius, and other pseudoplanktonic bivalves of different species are often fixed to floating ammonite shells (Seilacher 1982). Due to the strong compression of both sediment and ammonites, it is sometimes difficult to decide whether the bivalves originally grew upon the ammonite shell or were washed into the empty ammonite by currents or eaten by the ammonite or occupant of the empty ammonite shell. Only in those cases where small bivalve fragments are concentrated within the ammonite body chamber may these in fact represent food remains. In all 1 1 Harpoceras specimens, the bivalve debris forms but a single, usually longitudinal, rarely sub-circular, concentration 30-80 mm long and 20^40 mm wide in each specimen. The particle diameter is < 1-7 mm, rarely 10 mm. The fragments are mostly ribbed and originate either from Pseudomytiloides dubius , Pseudomonotis substriata (Munster), or Oxytoma inaequivalvis (J. Sowerby), well-preserved specimens of all three species being not rare within this horizon, and at least two of them had a pseudoplanktonic mode of life. In contrast with the crustacean remains, bivalve debris concentrations are positioned in sector 2 in all 11 specimens, often close to the ventral side (PI. 5, figs 3-4; PI. 6, figs 3-5), less often to the ventro-median or median side (PI. 5, figs 1-2; PI. 6, fig. 1). This different pattern, together with the total absence of aptychi in all 1 1 specimens and the presence of large washed-in shells in some of the specimens, requires a different interpretation of the bivalve debris. Because of its position in the anterior part of the body chamber, it certainly does not represent the stomach contents of the ammonite. In some specimens it may be the crop contents, especially in one of the two specimens where the bivalve debris is associated with crustacean remains, showing that portions of the ammonite’s soft parts were very probably still present in the body chamber. In the first specimen, the bivalve debris lies in between crustacean pereiopods, forming two balls in sector 2, median (crop content?). In the second specimen, it and the crustacean are found 70 mm apart. However, this is one of the few specimens where the food ball centres are positioned in sector 1 and probably are dislocated. In the other nine specimens with bivalve debris no crustacean remains occur. EXPLANATION OF PLATE 7 Figs 1^4. Remains of Coleial sp. and small aptychi in the body chamber of adult macroconchs of Harpoceras falciferum from the early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1, five food balls including remains of Coleial sp., plus small aptychi outside the food balls. The ammonite’s own large aptychus is also present. The sections enlarged in PI. 8, figs 1-3 are indicated by frames. W 21 (now in collection of Geo Centrum Brabant. Boxtel); x0-5. 2-4, two food balls including small aptychi, but no crustaceans. 2, W 35; the ammonite’s large own aptychus is indicated by an arrow. The sections enlarged in figs 3-4 are indicated by frames; xO-5. 3, x2-5. 4, x 3-2. PLATE 7 JAGER and FRAAYE, body chamber contents of Harpoceras 572 PALAEONTOLOGY, VOLUME 40 In the large ammonite Phylloceras heterophyllum (Sowerby) described and figured by Riegraf et al. (1984, p. 57, fig. 12) from the Unterer Schiefer of Dotternhausen, the shell debris of Pseudomytiloides dubius must represent crop content. In the Harpoceras specimens, however, the position of the bivalve debris probably reflects the position of the mantle cavity of the ammonite. As it seems improbable that a living ammonite was unable to remove debris from its mantle cavity, it may be that the debris was accumulated by a different animal (maybe crustacean, 'decapod kitchen’) living in the mantle cavity of a dead ammonite. The presence of soft parts in the dorsal region of the body chamber of the ammonite is needed to explain why the bivalve debris often is in a distinctly ventral position, although the ammonite was certainly buried lying flat on the sea-floor: in an empty body chamber the relatively heavy (in comparison with the crustacean remains) bivalve debris should flow to a median or even dorsal position, even if it was originally amassed in the ventralmost position at a time when the ammonite shell rested in a vertical position on the sea-floor. Specimens with small aptychi Aptychi much smaller than the aptychi of the ammonite in whose body chamber they are found are present in several of the Harpoceras falciferum with or without crustaceans, for example in a specimen with five crustacean balls (PI. 7, fig. 1 ; PI. 8, figs 1-3). If the small aptychi are complete, it cannot be determined whether they were eaten prey or were washed into the body chamber, for the body chamber often serves as a sediment trap containing complete ammonite shells up to several tens of millimetres in diameter. However, at least for a single Harpoceras specimen (PI. 7, figs 2^1), it appears that the small broken aptychi inside the body chamber come from prey either eaten by the ammonite (more probable) or by an occupant of the empty ammonite shell (less probable). It is an adult macroconch which preserves its large aptychus in the body chamber. There are two balls in sector 3 median: one contains three, the other at least nine small aptychi; the length of small aptychi is 3-10 mm, and of its own large aptychi 50 mm. Most small aptychi are complete; some, especially the 10 mm specimen, are somewhat broken. The balls also contain two siphuncles of ammonites of 3 and 5 mm diameter, as well as fine-grained detritus. Acknowledgements . We thank Mr Walter Bayer, Dotternhausen, for taking most of the photographs, and Dr John W. M. Jagt, Venlo, for improving our English. REFERENCES barthel, k. w. and janicke, v. 1970. Aptychen als Verdauungsriickstand. Ein Fund aus den Solnhofener Plattenkalken, unteres Untertithon, Bayern. Neues Jahrbuch fiir Geologie und Palaontologie, Monatshefte, 1970. 65-68. dagys, a. s., lehmann, u., bandel, k., tanabe, k. and weitschat, w. 1989. The jaw apparati of ectocochleate cephalopods. Palaontologische Zeit shrift. 63, 41-53. explanation of plate 8 Figs 1-3. Five food balls including remains of Coleial sp., plus small aptychi outside the food balls in the body chamber of an adult macroconch of Harpoceras falciferum from the Early Toarcian commune Subzone of Dotternhausen in south-west Germany. 1-3, W 21 (same specimen as PI. 7, fig. 1, now in collection of Geo Centrum Brabant, Boxtel); arrows indicate small aptychi. 1, x0-8; 2, x2-0; 3, x2 0. PLATE 8 JAGER and FRAAYE, body chamber contents of Harpoceras 574 PALAEONTOLOGY, VOLUME 40 fraaye, r. and jager, M. 1995. Decapods in ammonite shells: examples of inquilinism from the Jurassic of England and Germany. Palaeontology , 38, 63-75. haven, n. 1972. The ecology and behaviour of Nautilus pompilius in the Philippines. Veliger , 15, 75-81. jager, m. 1991. Lias epsilon von Dotternhausen, 2. Teil. Fossilien , 8, 33-36. 1993. Das Fossilienmuseum im Werkforum. Ein Fiihrer durch die Ausstellung von Jura- Fossilien. Rohrbach Zement, Dotternhausen, 128 pp. lehmann, u. 1972. Aptychen also Kieferelemente der Ammoniten. Paldontologische Zeitschrift , 46. 34-48. — 1975. Uber Nahrung und Ernahrungsweise von Ammoniten. Paldontologishe Zeitschrift, 49, 187-195. - — 1976. Ammoniten. Ihr Leben und ihre Umwelt. Enke, Stuttgart, 171 pp. — 1981. Ammonite jaw apparatus and soft parts. 275-287. In house, m. r. and senior, j. r. (eds). The Ammonoidea. Academic Press, London. — 1985. Zur Anatomie der Ammoniten: Tintenbeutel, Kiemen, Augen. Paldontologische Zeitschrift, 59, 99-108. — 1990. Ammonoideen. Leben zwischen Skylla und Charybdis. Haekel-Biicherei, Stuttgart, 2, 268 pp. — and weitschat, w. 1973. Zur Anatomie und Okologie von Ammoniten: Funde von Kropf und Kiemen. Paldontologische Zeitshrift, 47, 69-76. magnier, y. and laboute, p. 1978. Guides sous-marin de Nouvelle Caledonie. Les Editions du Pacifique, Paris, 160 pp. michael. r. 1894. Uber Ammoniten-Brut mit Aptychen in der Wohnkammer von Oppelia steraspis. Zeitschrift der Deutschen Geologischen Gesellschaft, 46, 697-702. morton, N. 1981. Aptychi: the myth of the ammonite operculum. Lethaia, 14, 57-61. — and nixon, m 1987. Size and function of ammonite aptychi in comparison with buccal masses of modern cephalopods. Lethaia , 20, 231-238. nixon, m. 1987. The diets of cephalopods, 201-219. In boyle, p. r. (ed. ). Cephalopod life cyles, 2. Academic Press, New York. — 1988. The feeding mechanisms and diets of cephalopods - living and fossil. 641-652. In wiedmann, j. and kullmann, J. (eds). Cephalopods - present and past. Stuttgart, 725 pp. riegraf, w., werner, G. and lorcher, F. 1984. Der Posidonienschiefer . Biostratigraphie , Fauna and Fazies des sudwest deutschen Untertoarciums ( Lias epsilon). Enke, Stuttgart, 195 pp. saunders, w. b. and landman, N. H. 1987. Nautilus. The biology and paleobiology of a living fossil. Topics in Geobiology, 6. Plenum Press, New York and London, xxih + 632 pp. seilacher, a. 1982. Ammonite shells as habitats in the Posidonia Shales of Holzmaden - floats or benthic islands? Neues Jahrbuch fur Geologie und P aldontologie , Monatshefte, 1982, 98-114. tanabe, k.., fukuda, Y., kanie, Y. and lehmann, u. 1980. Rhyncholites and conchorhynchs as calcified jaw elements in some late Cretaceous ammonites. Lethaia, 13, 157-168. tshudy, d. m., feldmann, r. m. and ward, p. d. 1989. Cephalopods: biasing agents in the preservation of lobsters. Journal of Paleontology, 63, 621-626. ward, p., stone, R., westermann, G. and martin, A. 1977. Notes on animal weight, cameral fluids, swimming speed, and color polymorphism of the cephalopod. Nautilus pompilius in the Fiji Islands. Paleobiology 3, 377-388. — and wicksten, m. 1980. Food sources and feeding behavior of Nautilus macromphalus. Veliger, 23, 119-124. ward, p. d. 1987. The natural history of Nautilus. Allen and Unwin, Boston, xiii + 267 pp. MANFRED JAGER Rohrbach Zement Fossilienmuseum D-72359 Dotternhausen Germany RENE FRAAYE Geo Centrum Brabant St. Lambertusweg 4 5291 NB Boxtel The Netherlands Manuscript received 13 February 1995 Revised manuscript received 18 November 1996 LATE DEVONIAN WINGED PREOVULES AND THEIR IMPLICATIONS FOR THE ADAPTIVE RADIATION OF EARLY SEED PLANTS by N. P. ROWE Abstract. Winged preovules, preserved as compressions, have been recovered from the upper Famennian of Sauerland in central Germany. Preovule organization of Warsteinia paprothii gen. et sp. nov. includes a nucellus with distal hydrasperman lagenostome and a preintegument comprising four alate lobes. The hydrasperman organization appears to be similar to that of other well-documented Late Devonian and Early Carboniferous preovules. The alate preintegument morphology is comparable to that of permineralized preovules, such as Lyrasperma scotica and Euryostoma burnense from the upper Tournaisian of Scotland. The preintegument is differentiated into a dense inner and fibrous wing-like outer zone. The Devonian preovules represent the earliest evidence of a winged plant diaspore, and the differentiation of a preintegumentary sclerotesta and sarcotesta could be regarded tentatively as an adaptation to optimize wind-mediated dispersal. The variety of integument structures following the earliest preintegument suggests that the system of unfused, terete preintegument lobes among earliest seed plants represented an important pre-aptation for a range of functions including protection, optimization of pollination, and dispersal. The appearance of the seed habit among Palaeozoic lignophytes represented one of the most important evolutionary innovations in land plants. Much has been discussed concerning the evolutionary significance of the seed habit, its pattern and timing of appearance, and its adaptive significance in conferring a number of potential selective advantages (Long 1960a, 1975; Andrews 1963; Pettitt and Beck 1968; Gillespie et al. 1981; Matten and Lacey 1981; Niklas 1981a, 19816, 19836, 1985; Tiffney 1986; Rothwell and Scheckler 1988; DiMichele et al. 1989; Haig and Westoby 1989; Rowe 1992a). The earliest seed plants differed radically from plants with ‘ free-sporing’ pteridophytic reproduction, in which sporangia were dehiscent and spores containing the haploid gametophyte generation were released directly into the environment. Under these circumstances, gametophyte growth, development of ova and spermatozoids, and syngamy (fertilization) all occurred unprotected from and unprovisioned by the parent plant sporophyte. The earliest seed plants of the late Devonian and early Carboniferous show two basic structural changes of the megasporangial complex which vastly modified sexual reproduction (Text-fig. 1). Firstly, the megasporangium no longer simply comprised a relatively uniform wall structure capable of simple dehiscence and functioning to disseminate spores, but possessed an apical opening (lagenostome) and differentiated ‘pollen chamber’, which functioned to trap and then direct wind-borne microspores towards the close proximity of the female gametophyte. Secondly, the megasporangium, or ‘nucellus’ as it is termed in seed plants, was surrounded by a layer of sterile tissue composed of four or more vascularized lobes known as the preintegument. This system of lobes is generally fused at the base, and either fused or adpressed to the surface of the nucellus inside. With the appearance of the seed habit, the development of the female gametophyte, pollination and possibly fertilization all took place within a complex structure still attached to and provisioned by the parent plant sporophyte. Generally accepted terminology refers to the unfertilized nucellus/preintegumenl complex as a preovule. Once fertilization has occurred and an embryo is present, the structure is referred to as a seed. The terms ‘preovule’ and ‘preintegument’ are currently used to distinguish a morphological | Palaeontology, Vol. 40, Part 2, 1997, pp. 575-595, 3 pis) © The Palaeontological Association 576 PALAEONTOLOGY, VOLUME 40 text-fig. 1. Longitudinal section through the middle of a generalized archetypal ovule (unfertilized seed) of an early seed plant. The haploid megagametophyte (mg) is enclosed within a single functional megaspore and surrounded by the megaspore membrane (mm). The megaspore is retained within a modified mega- sporangium known in seed plants as the nucellus (n). Pollination is effected via the specialized apex of the nucellus. Pollen enters the nucellar apex through the distal aperture known as the lagenostome (la) and is retained just above the female gametophyte in the pollen chamber (pc). Further development of the ovule, involving growth of the megagametophyte is believed to rupture the ‘floor’ of the pollen chamber and thus bring the microgametophyte into direct contact with the megagametophyte and archegonia (ar) resulting in fertilization. The central column (cc) situated in the pollen chamber is believed to enlarge and block the distal opening after fertilization. The nucellus and megagametophyte are surrounded by a ring of four or more slender pointed lobes (pi) known as the preintegument. In this diagram the section has passed along the middle of the lobes on the left and the right of the ovule and a total of six lobes would be expected. The preintegument typically has vascular strands (vs) of conducting tissue which are continuous with the parent plant via the pedicel (pe). In many of the earliest seed plants, the ovules are inserted in branched vegetative structures known as cupules (not shown). Scale bar represents 1 mm. grade of development among seed-plants in which the sterile lobes surrounding the nucellus are not entirely fused. This differs from the arrangement generally observed in practically all seed plant ovules by the end of the Carboniferous, when the integument formed an entire structure around the apex of the nucellus, surrounding a space known as the micropyle and replacing the pollen-trapping function of the lagenostome. Whilst there is general agreement that the seed habit represented an innovation in terms of the overall reproductive biology of land plants, there has been more uncertainty concerning the precise functional and selective significance of the principal morphological structures of the preovule. Central to this debate has been the functional role(s) of the preintegument, and its adaptive significance in terms of protection of the megagametophyte and embryo, and optimization of the aerodynamic properties of the preovule to enhance pollination. The traditional interpretation stresses a protective role of the preintegument such as a defence against dehydration, herbivory and pathogens. An alternative viewpoint, suggested by empirical biomechanical experiments, suggests that the morphology of the preintegument may have undergone modification as a result of selection for optimization of the aerodynamic properties of the megasporangium and the enhancement of pollination from wind-borne pollen (Niklas 1981a, 1983a, 19836, 1985). Although both hypotheses may broadly contribute to an understanding of the potential ecological and evolutionary scenarios concerning the early evolution of seed plants, there is insufficient evidence to state whether one or the other functions of the preintegument represents an adaptation in the strict sense (Gould and Vrba 1982) or several pre-aptive functions of the preintegument. An explicit determination of the adaptive significance of the preintegument is confounded by the absence of a rigorous and sufficiently inclusive phylogenetic history of early lignophytes (progymnosperms and seed plants). ROWE: DEVONIAN WINGED PREOVULES 577 It is therefore unknown whether the preintegument, initially manifested as a ring of preintegu- mentary lobes, appeared before or after morphological differentiation of the lagenostome (the elaborate opening of the nucellar apex) and it is probably premature to interpret the adaptive significance of the characteristic terete preintegumentary lobes characterizing many of the earliest gymnosperms as either resulting from selection for protection or optimized pollination. The difficulty of attempting ecological and/or functional interpretations of the preintegument has been demonstrated recently by the discovery in sedimentary rocks post-dating the earliest Devonian preovules of a preovule lacking a lagenostome at the nucellar apex but possessing well-developed preintegument lobes (Galtier and Rowe 1989, 1991; Rothwell and Serbet 1992; Bateman and DiMichele 1994). Furthermore, a putatively ‘ancestral’ megasporangium lacking a preintegument, but equipped with a distal aperture, might, from extrapolation of the aerodynamic properties of an entirely integumented ovule, be more efficient than a partially entire or lobed preintegument. This calls into question whether the appearance of the preintegument could be interpreted credibly as an adaptation to aid pollination (Haig and Westoby 1989). Many authors have discussed ecological and evolutionary scenarios based on the documented range of preovule morphologies, particularly in terms of preintegument morphology, from the upper Tournaisian and upper Visean of Scotland. Generally speaking, this range of forms includes those with slender unfused preintegument lobes (morphologically similar to the earliest seed plant preovules) such as Genomosperma (Long 1959) and those with an entire integument such as Stamnostoma (Long 1960/?). This contemporaneous morphological diversity has been seen as possible evidence that early seed plants went through a period of n-selection (Arthur 1984, 1988) of low interspecific competition in empty ecospace (DiMichele et al. 1989) or that differing preintegument morphologies reflected a range of possibly different habitats (Niklas 1992, p. 521). Both are attractive though contrasting scenarios for explaining the morphological variation of the preintegument during the early Carboniferous, preceding the general observed trend of lobe fusion and integument-delimited micropyle towards the end of the Carboniferous. Knowledge of the earliest seed plants has increased rapidly in the last decade as a result of intensive investigations of ancient seed plants from three principal late Devonian fossil localities in Europe and North America. These include the Elkinsia assemblage, Hampshire Formation in West Virginia (Gillespie et al. 1981 ; Rothwell and Scheckler 1988; Rothwell et al. 1989); the Moresnetia assemblage from the Evieux Formation in Eastern Belgium (Fairon-Demaret and Scheckler 1987) and the Laceya assemblage from the Coomhoola Formation, Ballyheige in south-west Ireland (Matten et al. 1980a, 19806, 1984). All contain preovules with a relatively uniform morphology, consisting of a nucellus with an apical lagenostome enveloped by a preintegument comprised of four to ten slender, terete lobes. In all three cases the preovules are inserted within cupules consisting of slightly flattened segments which are divided distally into narrower lobes. The main source of variation mostly concerns differences in size and number of preintegument lobes, and their degree of fusion with each other and the nucellus, as well as differences in symmetry, branching and organization of the sterile units comprising the cupule. At present, Elkinsia polymorpha , Rothwell, Scheckler and Gillespie, 1989 is probably the most completely reconstructed early seed plant, but much information is required to reconstruct most of these plants completely. Hypotheses have been put forward concerning their likely habitats in marginal, ephemeral situations with an ecological strategy as rapid colonizers of new emergent habitats. These scenarios intuitively embrace the selective advantages of pollination, protection and dispersal resulting from acquisition of the seed habit. Although the fossil record of the earliest seed plants is well documented from upper Famennian clastic sequences of Euramerica, marine sediments dominate many areas directly succeeding the Devonian/Carboniferous boundary. It is only by the upper Tournaisian that plant assemblages are known demonstrating diverse anatomically preserved spermatophyte fertile structures, and which reflect morphological radiation on a large scale with significant structural and functional divergences from the ‘ancestral gymnosperms’ of the late Famennian (Long 1960a, 19606, 1969, 1975, 1977a, 19776). Many of the putative seed plant findings from uppermost Devonian, and lower 578 PALAEONTOLOGY, VOLUME 40 and middle Tournaisian sedimentary rocks are known only from marine or highly allocthonous deposits where seed plant reproductive organs are uncommon and/or poorly preserved. Recent findings from the middle Tournaisian ‘Lydiennes’ deposits of southern France exemplify this situation with a single specimen of Coumiasperma remyii recorded (Galtier and Rowe 1989, 1991). More articulated compression material from a lithological variation of the ‘ Lydiennes’ at La Serre, southern France, has yielded a variety of putative seed plant, rachises, foliage and cupules but lacking unequivocal in situ preovules (Rowe and Galtier 1990). While the morphology of some of these cupule compressions may be consistent with those seen in late Famennian gymnosperms others are reminiscent of more derived, compact, cupules from the Lower Carboniferous of Scotland. If the interpretation of the compression material from La Serre is correct, it provides compelling evidence that significant diversification of cupule morphology had occurred by the mid Tournaisian. The object of this paper is to describe and discuss some preovules with winged preintegumentary lobes from the upper Fammennian of central Germany. Although, only occurring as isolated units, they provide evidence of significant diversification of preovule morphology prior to the end of the Devonian and the wide morphological diversity observed in the late Tournaisian. LOCALITY INFORMATION The Hangenberg Sandstein at Oese, central Germany is situated in an abandoned roadside quarry beside the B7 road, between Menden and Hemer, approximately 12 km south of the river Ruhr (Higgs and Streel 1984; Keupp and Kompa 1984). At the base of the section, Wocklum limestone is succeeded by a narrow bed of Hangenberg Schiefer which is typical of the Upper Devonian lithofacies of the Remscheid-Altena anticline in the Rheinisches Schiefergebirge (= Rheinish Slate Mountains) of central Germany (Higgs and Streel 1984). At Oese, a local variation of the Hangenberg Schiefer is present as a series of coarser sandstones with a south-westerly provenence and interpreted as an off-shore deposit in a high energy depositional environment in the proximity of off-shore reefs (Keupp and Kompa 1984). At Oese, approximately 12 m of sandstones are succeeded by a thin band of Hangenberg limestone which marks the Devonian/Carboniferous boundary (Luppold et al. 1994). Beds 2-4 m below the D/C boundary in the Hangenberg Sandstein have been dated stratigraphically on the basis of conodont faunas giving a lower Protognathodus Zone, and meiospore assemblages giving a LN biozone in the uppermost Famennian. Above the boundary beds, Tournaisian sedimentary rocks are dominated in the region of the Rheinish Slate Mountains by black marine shales (Alaunschiefer) and radiolarian cherts which are also exposed in a small subsidiary cutting adjoining the main quarry at Oese. Although permineralized plants have been identified from these sedimentary rocks and from a nearby locality at Oberrodinghausen (Rowe 1992 b\ Rowe et al. 1993), plant material has not been recovered from the Alaunschiefer at Oese. Plant fossil material is found throughout the exposed sequence of Hangenberg Sandstein. The sandstone consists of at least eight to twelve beds up to 1 m thick consisting of upward fining sequences from very coarse, micaceous, sandstones to narrow layers of fine shale. The plant macrofossils may be divided into four broad categories, which appear to be facies controlled and comprise: (1) poorly preserved, flattened compressions of axes (sometimes orientated) up to 015 m long, recovered from coarse micaceous levels throughout the section; (2) three-dimensional moulds and casts of axes most usually confined to the coarses facies; (3) dark red to light orange, petrified, (probably limonitized) axes up to 0 35 m long showing anatomy and found throughout coarse sandstone and fine shale bands; (4) highly fragmentary, densely deposited, plant meso-debris, occurring in thin bands (10^10 mm) of laterally discontinuous, localized shales at the top of several upward fining sequences of coarse sandstone. The preovules come predominantly from the type (4) facies association. They are similar in size to much of the rest of the indeterminable plant meso-debris and are the most common identifiable plant organs. The remainder of the meso-debris consists of indeterminable axes, rarely more than ROWE: DEVONIAN WINGED PREOVULES 579 text-fig. 2. a-k, preservational variants of Warsteinia paprothii. The ovule compressions vary considerably in the appearance of the nucellar apical region, in the appearance and integrity of the outer preintegument lobe tissue and in the variable positioning and appearance of medianly positioned preintegument lobes. a, V. 64114a; b, V. 641 15a; c, V. 64116a; d, V.641 17a(l); e, V.641 18a( 1 ); f, V. 64119; G, Vfi4118a(2); h, V. 64120; i, V.641 1 8a( 3) ; j, V.641 121a; k, V.641 13a. Scale bar represents 1 mm. 1-2x10-20 mm, megaspores and fragments of highly divided foliage and broader spatulate pinnules. Exceptions to this broad category include a branched non-cupulate ovulate structure and rare ‘ leaf’ fragments with an open dichotomous nervation which are comparable to compression foliage of Archaeopteris. Small fragments of fossil charcoal (pyrofusain) are also consistently observed in this facies. Plant material from the coarser facies type consists of anatomically preserved axes of cladoxylaleans and protostelic lignophytes reaching up to 0 35 m long. Also common are compressions and casts of petiolate, bifurcate rachises, occasionally with pinnules in attachment, and compressions and casts of leafy or decorticated lycopsid axes, microphylls and sporophylls. METHODS AND MATERIALS Preovules are preserved as carbonaceous compressions throughout the exposed section of the Hangenberg Sandstein, predominantly from the finer-grained horizons and occasionally from the coarser, highly micaceous sandstones. Most specimens show evidence of organic material visible as black vitrinite or dark to light brown material resembling oxidized cuticle. Oblique, semi-polarized light and SEM observation of the finest impression surfaces occasionally yielded limited data on cell size and alignment. Uncovering the material with fine needles was successful in revealing substantial parts of the preovule compression but smaller-scale uncovering was difficult because of organic material adhering to the large sediment grains and mica. Interpetation of small-scale three-dimensional features involved critical observation of the course of the surfaces exposed by the fracture plane through the compression fossil. Embedding followed by sectioning with a diamond wafering blade did not yield informative sections. The three dimensional arrangement of the preovule was made possible by careful consideration of both parts and counterparts of a number of key specimens. Macrophotography (up to x 16) was carried out on a Zeiss Tassovar with intense, polarized, fibre optic light source. Higher magnification was 580 PALAEONTOLOGY, VOLUME 40 achieved by placing the specimen directly on the stage of a Nikon compound microscope fitted with incident cross-polarized lighting and Nikon BD Plan objective lenses. This technique provided high magnification micrographs up to x 230, rivalling those obtained from cellulose acetate surface peels but leaving the compression intact. All specimens are deposited in the Department of Palaeontology, The Natural History Museum, London, specimen numbers V. 641 13 to V. 64121. DESCRIPTION More than 160 isolated preovules were examined. There is a high degree of variation in shape, overall size and presence or absence of main features. Study was confined to well-preserved specimens showing combinations of at least two of the following structures: (1) a preintegument, (2) a nucellus, (3) apical elaboration of the nucellus into a lagenostome. Based on these criteria, approximately 20 specimens showing well-preserved, key characters were selected for study. Many other specimens from the assemblage lacked one or more of these structures and therefore could not be interpreted definitely as preovules. P re ovule morphology The most complete preovule compressions have overall dimensions varying from 3-3-4-7 mm long and 2-3—3 0 mm wide (Text-fig. 2). The preovules consist of a central oval nucellar region, a distal extension of this into an apical langenostome, and a preintegument. The appearance of all three of these structures is strongly dependant on the three-dimensional extent of the compression, the degree of sediment accretion around the lagenostome and preintegument lobes, and the passage and direction of the plane of fracture passing through or around the compression/matrix structure when the rock was split. The nucellar region is pointed proximally and broadens to 1 - 1—1-4 mm just above the mid-level of the preovule (PI. 1, figs 1-2). Near to the apex of the preovule, the nucellus is contracted slightly (PI. 2, figs 1-2) and this level is interpreted as the distal limit of the pollen chamber. In some specimens the structure interpreted as the lagenostome is visible as an extension of the nucellar apex, about 05 mm wide and 05 mm long. The original 'tube-like' nature of the lagenostome is visible in some specimens where sediment has entered the opening and lagenostome has retained a three- dimensional structure as a minute cast, in contrast to the rest of the preovule which became flattened (PI. 2, fig. 2). The appearance of the preintegument is the most variable feature as a result of a variety of presumed taphonomic processes. Well-preserved, laterally orientated compressions often show evidence of a preintegumenl lobe at each side of the nucellar region (PI. 1, fig. 1 ; PI. 2, fig. 3). Each EXPLANATION OF PLATE 1 Figs 1-5. Warsteinia paprothii gen. et sp. nov.; Oese; Upper Devonian. 1, V. 641 13a, holotype (part); carbonaceous compression of entire winged ovule showing evidence of two prominent alate preintegument lobes to the left and right and a lagenostome at the apex of the nucellus; x 25. 2, V.641 14a; ovule with slender nucellar region and highly abraded alate preintegument lobes; a common preservational variant at the locality, in which alate ovules resemble ovules with slender preintegument lobes; evidence of three preintegument lobes is visible on the part of this specimen; the lagenostome is well preserved and sediment filled; x25. 3, enlargement of fig. I ; fimbriate appearance of the alate preintegument lobe compression consisting of an inner dense region (lower right) and an outer region of ‘radiating strands’; x 60. 4, enlargement of fig. 1 ; preintegument lobe showing continuity of less-dense organic compression material between ‘radiating strands’; x 60. 5, V.641 17a(2), preintegument lobe of organically well-preserved ovule showing continuity between radiating strands and integrity of the outer margin; x 50. PLATE 1 ROWE, Warsteinia 582 PALAEONTOLOGY, VOLUME 40 A B text-fig. 3. Interpretive diagram of three-dimensional disposition of preintegument lobes from part and counterpart of single specimen of Warsteinia paprothii, V.641 17(1), (see also PI. 3). a, V.641 1 7a( 1 ) ; ovule part with evidence of preintegument lobes (1-3); the central lobe (2) is not exposed by the plane of fracture but is visible as a longitudinally aligned groove filled with organic material, b, V.641 17b( 1), counterpart of ovule also showing parts of lobes 1 and 3 in addition to the medianly positioned lobe 4; the plane of fracture has removed the sediment separating the distal parts of lobes 1 and 4 to expose the compression surface of lobe 4. c-d, idealized section through distal (c) and proximal parts (D) of part/counterpart complex. lobe consists of two discrete zones; a dense inner organic layer, and an outer network of interconnected but generally radially aligned strands (PL 1, figs 3-5). These are referred to as the sclerotesta and sarcotesta respectively. Each preintegument lobe is 065-077 mm wide at its broadest point just below the mid-level of the preovule, and tapers gradually to the apex about 0-6-08 mm above the lagenostome. Around the base of the preovule the lobe is only slightly narrower than at the widest part. Differentiation of the challazal region is unclear but several specimens (e.g. PI. 3, figs 1-2) indicate a slender proximal extension of the nucellar region resembling part of a stalk. Preintegument lobes are either fused or consistently adpressed to the nucellus for two-thirds of the total preovule length, and appear to separate from the nucellus just below the level of the lagenostome and the inferred position of the pollen chamber (PI. 1, figs 1-2). The darker sclerotesta consists of much denser organic material, which is comprised of longitudinally aligned strands, or cellular elements (PI. 2, fig. 4) which taper distally but remain conspicuous above and below the level of contact of the preintegument to the nucellus (PI. 1, fig. 1). The sarcotesta occupies the greater part of each preintegument lobe and comprises less dense, generally radially aligned fibre-like tissue, which forms an interconnecting network (PI. 1, figs 3-5). ROWE: DEVONIAN WINGED PREOVULES 583 This is delimited on the outside of the preintegument lobe, in exceptionally preserved specimens, by a peripheral continuation of the fibre-like network forming the outer perimeter of each ‘wing’. This has a highly variable appearance (PI. 1, figs 3-5), according to preservational differences between specimens, and varies from a simple fimbriate appearance (PI. 1, fig. 3), to one in which continuous organic material is visible between fibrils (PI. 1 , fig. 4), to one in which a continuous outer perimeter is organically preserved (PI. 1, fig. 5). Individual strands are 15-40 /tm wide and anastomose, undergoing from two to four divisions to the outside of the sarcotesta. They are interconnected laterally by less dense organic material, which apparently integrades in density between adjacent strands. Organization of the preintegument Few specimens show direct evidence of the number of preintegumentary lobes. Evidence of more than two lateral preintegument lobes is variable, and depends on the degree of three dimensional preservation and orientation of the preovule in the matrix. Highly flattened specimens may show one or two darker carbonaceous longitudinal bands along the mid-region of the nucellus (Text-fig. 2). More than two lateral preintegument lobes are seen most clearly among the rarer three- dimensionally preserved preovules where they are visible as prominent longitudinal grooves (PI. 3, figs 1-4; Text-fig. 3). Occasionally the plane of fracture follows one or more of these grooves to expose the planar surface of a third preintegument lobe beneath (PI. 3, fig. 4, arrow; Text-fig. 3b, lobe 4). Text-figure 3 depicts part and counterpart of a three-dimensionally preserved preovule in which substantial sediment accretion occurred between preintegument lobes during burial of the seed and prior to compression (PI. 3, figs 1-2). Both part and counterpart show a preintegument lobe visible as a longitudinal groove along the mid-region of the nucellus, in addition to the pair of lobes at each side (Text-fig. 3a-b). The preintegument lobes are numbered 1^1 in an anti-clockwise direction when part and counterpart are depicted together (Text-fig. 3c-d). The plane of fracture has exposed all of lobes 1 and 3 but has only exposed the distal part of lobe 4, which is visible proximally as a longitudinal groove. The new structures are interpreted as seed plant preovules, as a result of the identification of a preintegument divided into lobes surrounding a nucellus with a differentiated apical lagenostome. The basic organization of this morphology is consistent with that of other well-documented preovules from the upper Famennian to the Upper Carboniferous. The organization of the sclerotesta and sarcotesta suggests that the former is entire and fused, or adnate to the nucellus to just below the base of the lagenostome. It is not possible to determine the precise extent of fusion between preintegument lobes. The density of the organic material remaining around the long axis of each preintegument lobe and around the nucellus suggests that the sclerotesta is continuous or fused around the nucellus. In each preintegument lobe, the sarcotesta is differentiated into a winged extension. Above the apical part of the nucellus the preintegument lobes arch out slightly above the level of fusion of the preintegument with the nucellus. Nothing is currently known about the mode of attachment of the preovule or whether the preovules were cupulate. SYSTEMATIC PALAEONTOLOGY Division spermatophyta (following Rothwell and Serbet 1994) Order lyginopteridales? (sensu Barnard and Long 1975) Family unknown Genus warsteinia gen. nov. Derivation of name. Warstein, a town in Sauerland, Germany. Type species. Warsteinia paprothii sp. nov. from the upper Famennian, Oese, Sauerland, Germany. 584 PALAEONTOLOGY, VOLUME 40 Diagnosis. Compressions of isolated preovules with four winged preintegumentary lobes. Preovules 3-3-4-1 mm long and 2-3-3-0 mm wide. Nucellar region oval, 11—1-4 mm wide in broadest part at mid-level of preovule, pointed proximally and differentiated distally into apical lagenostome, 0-5 by 0-5 mm, with angular proximal differentiation of presumed pollen chamber. Preintegument lobes adnate or fused to nucellus to just below pollen chamber, comprising an inner dense sclerotesta composed of longitudinally aligned elements and an outer sarcotesta comprised of perpendicularly aligned anastomosing fibrous strands. Warsteinia paprothii gen. et sp. nov. Plates 1-3; Text-figures 2^4 Derivation of name. In honour of Dr Eva Paproth. Holotype. V.64113a-b; roadside quarry on the B7 road at Oese between Menden and Hemer, Sauerland, central Germany; Hangenberg Sandstein, Upper Devonian, upper Famennian, LL-LN miospore biozone. Diagnosis. As for genus. This is the only recognized species. COMPARISONS WITH OTHER DEVONIAN AND EARLY CARBONIFEROUS PREOVULES Comparisons of other late Famennian preovules indicate that, apart from differences in overall size, and number and degree of fusion of preintegument lobes, there is little major difference in morphology (Fairon-Demaret and Scheckler 1987; Rothwell and Scheckler 1988; Rothwell et al. 1989). The best known Famennian preovules ( Elkinsia polymorpha Rothwell, Scheckler and Gillespie, 1989, Moresnetia zalesskyi Stockmans, 1948 and Kerryia mattenii Rothwell and Wight, 1989) all consist of a preintegument divided into four or five, or eight to ten lobes surrounding a nucellus differentiated apically into a pollen chamber and lagenostome. Differences in preovule morphology mainly concern the number of preintegument lobes and the degree of fusion or adpression with each other and the nucellus. Preintegument lobes are fused to the nucellus for one- third and one-half the length in Elkinsia and ovules of Kerryia, respectively, but only at the base in Moresnetia. This appears to be the most marked difference separating these preovules. Other differences include the relative size and shape of the pollen chamber and lagenostome, and the overall size and shape of the preovule. However, these characters are difficult to quantify between genera due to presumed changes associated with ontogeny of the gametophyte, pollination, fertilization, and sporophyte development. The main characters distinguishing the German EXPLANATION OF PLATE 2 Figs 1-4. Warsteinia paprothii gen. et sp. nov.; Oese; Upper Devonian. 1, V.641 13a, holotype; enlargement of Plate 1, fig. 1, showing continuity of preintegument lobe with nucellar region to just below the lagenostome, and the angular differentiation between the top of the nucellus (probable pollen chamber) and the lagenostome (arrow); x 42. 2, V.641 14a; enlargement of Plate 1, fig. 2, showing similar differentiation at the apex of the nucellus as seen in Plate 2, fig. 1; the remains of the highly abraded alate lobe are just discernible (arrow a), while a linear groove clearly marks the position of a third preintegument lobe in the median part of the nucellar region (arrow b); x 47. 3^1, V.641 20. 3, proximal part of an ovule exposed by the plane of fracture which has removed the apical part of the nucellus and lagenostome (a common preservational variant); x25. 4, enlargement of fig. 3 showing evidence of longitudinally orientated carbonaceous material (inner preintegument) continuous with perpendicularly orientated alate part of the outer preintegument; x 44. PLATE 2 ROWE, Warsteinia 586 PALAEONTOLOGY, VOLUME 40 preovules from Elkinsia, Moresnetia and Kerryia are the winged preintegument lobes differentiated into sclerotesta and sarcotesta. In surface view, each winged lobe occupies up to 60 per cent, of the area of the compressed nucellar surface. The preintegument lobes of the new seeds are over twice the width calculated from published accounts of Moresnetia and Kerryia (Fairon-Demaret and Scheckler 1987; Rothwell et al. 1989). In addition to this, there is no evidence in any of the genera mentioned above of a preintegument lobe organization more complex than a vascular strand positioned centrally in a simple, terete, distally tapered lobe (Text-fig. 1). In Archaeosperma arnoldii Pettitt and Beck, 1968 the preintegument lobes are fused up to the level of the pollen chamber area, above which each extends as a small terete tip. The surface of the preintegument of Archaeosperma possesses spine-like emergences which can appear quite dense in laterally orientated preovule compressions. These appear similar to poorly preserved preintegument lobes of Warsteinia in cases when the outer margin of the sarcotesta is not visible and the thinner organic material grading between the radially orientated strands is not preserved (PI. 1, figs 2-3). Xenotheca devonica Arber and Goode, 1915 from the upper Famennian of north Devon is one of the least known cupulate preovules (see also Rogers 1926). More recent reports of these structures (Fairon-Demaret and Scheckler 1987; Rothwell and Scheckler 1988) indicated that the terete, apparently slightly fused preintegument lobe organization of the preovules, and the morphology of the cupules, is similar to Elkinsia and Moresnetia. One of the least-well understood putative preovules from the Upper Devonian is Spermolithus devonicus Chaloner, Hill and Lacey, 1977 from the ‘Old Quarry' at Kiltorcan in southern Ireland. The outline and shape of these seed-like structures were compared with compressions of platyspermic seeds assigned to the form genus Samar opsis and the anatomically preserved Lyrasperma scotica (Calder) Long, 1960c/ from the Lower Carboniferous of Scotland (Calder 1938; Long 1960a). Morphological evidence was considered sufficient to interpret the Kiltorcan structures as platyspermic, bilaterally symmetrical gymnospennous preovules. There is apparently no evidence of an elaborated nucellar apex in the central, apical position of Spermolithus which would be expected if the material was of the same structural complexity as Lyrasperma (Stewart 1983, p. 236) or indeed any other preovule with a hydrasperman modification of the nucellar apex. The main question is whether both an integument and a nucellus are present in the material. Chaloner et al. (1977) interpreted a central, oval structure in the centre of each structure as a megaspore. The boundary between this and the outer surrounding structure is relatively clear, but outside the central oval area there is no consistent differentiation suggesting the presence of both an integument and a megasporangium/nucellus. It is therefore possible that the material represents a megaspore within a dehiscent megasporangium, as suggested perhaps by the irregular opening or split at the apex of the structure. This interpretation is perhaps more likely in the absence of convincing evidence of an integument and nucellus, or a structure resembling an elaborated nucellar apex or lagenostome. The preovules from Oese clearly differ from Spermolithus in possessing features consistent with a hydrasperman preovule morphology. In addition, the platyspermic appearance of certain preovule compressions EXPLANATION OF PLATE 3 Figs 1-4, Warsteinia paprothii gen. et sp. nov.; Oese; Upper Devonian; part, V.64 1 1 7a( 1 ) and counterpart, V.641 1 7b( 1 ) of single ovule showing preservational organization of four preintegument lobes (see also Text- fig. 3). 1, V.641 17a(l), showing evidence of three preintegument lobes (arrows), lobe 1 at left, lobe 3 at right and lobe 2 towards the right of the nucellar region; x 28. 2, V.641 1 7b( 1 ) counterpart, showing evidence of preintegument lobes 1 and 3, to the right and left of the specimen; a fourth preintegument lobe (lobe 4) is visible towards the right of the nucellar region ; x 44. 3, enlargement of fig. 1 , showing carbonaceous material of lobe 2 disappearing into the matrix; lobe 2 is separated from lobe 3 by a linear sliver of fine matrix (arrows); note the attenuated proximal outline of the ovule suggestive of differentiation into a narrow basal stalk; x 28. 4, enlargement of fig. 2, showing where the plane of fracture has exposed the distal surface of preintegument lobe 4 (arrow) by passing through the sliver of sediment separating lobes 1 and 4; x 44. PLATE 3 ROWE, Warsteinia 588 PALAEONTOLOGY, VOLUME 40 may be belied by the effects of compression. This is clearly the case with Warsteinia in which four preintegument lobes are present with a probable rotational symmetry (Rothwell 1986) of 90°. Well-documented preovules from the Lower Carboniferous are known mostly from strata no older than the CM biozone of the upper Tournaisian. Many anatomically preserved, putative seed plants from earlier in the Tournaisian are found only in marine black shales and cherts. These sedimentary rocks yield mostly vegetative stems, petioles and rachises, which were probably rafted into the marine depositional environment, and contain very few reproductive organs. As a result, very few seed or cupulate organs are known from this period so that much of our knowledge of the diversity of early seed plants in the earliest Carboniferous, is limited to vegetative axes and leaves. One single example of an anatomically preserved preovule is known from the middle Tournaisian ‘Lydiennes’ formation of the Montagne Noire in southern France (Galtier and Rowe 1989, 1991). The structure consists of eight relatively massive preintegument lobes surrounding the nucellus. The apex of the nucellus lacks a pollen chamber and consists only of a solid parenchymatous nucellar beak, in spite of the fact that preovule ontogeny had proceeded sufficiently to develop cellular endosporic gametophyte tissue. The pattern observed in the compressed preintegument lobes of Warsteinia could conceivably represent, in terms of size and width to the nucellus, a compressed version of the permineralized preintegument lobes of Coumiasperma. However, the aerenchymatous tissue and position of the vascular trace in Coumiasperma is not consistent with the arrangement of inner dense and outer strand-like organic material observed in Warsteinia. Firstly, in Coumiasperma the vascular strand is positioned centrally in the preintegument lobe. Secondly, the tissue of the preintegument lobe is aerenchymatous and, although some of the cellular components are aligned perpendicularly, there is no evidence of well-defined fibrous elements. More than 25 species from 15 genera of permineralized preovule are known from the upper Tournaisian to upper Visean of Scotland (Rothwell 1986). Apart from the record of Coumiasperma from the middle Tournaisian of France, the assemblages from Scotland have, up to now, represented the earliest evidence of a wide-scale diversification in preovule morphology, which is observed mostly in terms of preintegument structure with differences in size and shape of the lagenostome and pollen chamber relative to the nucellus. Apart from size, structural variation of the preintegument may be characterized broadly among Early Carboniferous preovules in terms of: (1) number and symmetry of preintegument lobes; (2) extent of fusion or adpression of the preintegument to the nucellus; (3) position of vascular strand in each preintegumentary lobe; (4) presence/absence of hairs, spines; and (5) lobe/ridge differentiated into endotesta (sclerotesta) and sarcotesta. The well-documented preovules from the Famennian show variations in some or all of characters (1) to (5). Some of the preovules from the CM biozone and younger strata, such as Hydrasperma tenuis Long, 1961, differ so little from the ‘ancestral’ preovule morphology described for Elkinsia , Moresnetia and Kerryia , that as isolated units they cannot be distinguished from the older preovules (Matten et at. 1980a; Rothwell and Wight 1989). The fifth main type of divergence is the appearance of a ‘winged’ preintegument as shown by Lyrasperma scotica and Euryostoma burnense (Long) Long, 1975. Both preovules possess a preintegument differentiated into sclerotesta and sarcotesta, the latter forming extended ‘wings’ or ‘keels’. The inner integument is entire around the preovule proximally, but more distally both endotesta and wing-like sarcotesta are individual free winged lobes (Calder 1938; Long 1960a, 1969). It is this organization which is most comparable to that observed in the new Famennian preovules from Germany. Winged lobes are most developed in Lyrasperma. In E. angulare Long, 1969, there is a broadly similar differentiation of inner and outer preintegument tissue but the lobes are narrower (Calder 1938; Long 1960a). In E. burnense the inner sclerotesta consists of dense, longitudinally aligned fibrous tissue material completely surrounding the nucellus, which also partially envelops the vascular bundle of each winged lobe (Long 1969). Both the sclerotesta and vascular bundles are positioned close to the inner surface of the preintegument in contact with the nucellus. In E. burnense the tissue outside the sclerotesta is described as being comprised of alternating bands or plates of thick-walled cells with softer tissue which seemed to have decayed. ROWE: DEVONIAN WINGED PREOVULES 589 In Lyrasperma a similar but not identical organization of tissue types is observed in which the vascular bundles of each preintegument lobe are partially enveloped by radially aligned flanges of the inner sclerotesta. The wing-like extensions forming each preintegument lobe are comprised of transversely aligned, multicellular strands apparently separated by intercellular spaces (Calder 1938; Long 1960a). Warsteinia paprothii is smaller than all three preovules discussed above by factors of four to eight. Because of differences in preservation it is difficult to make detailed comparisons. There is a strong implication that the winged preintegument organization comprises a dense inner tissue possibly containing the preintegument vascular strand (Text-fig. 3) and an outer wing-like tissue comprised of resistant fibrous tissues alternating with thinner material. DISCUSSION The preovules from Germany (Text-fig. 4) demonstrate that significant divergence in preintegument morphology had occurred among seed plants with hydrasperman reproduction before the end of the Devonian. This obviously predates the late Tournaisian, CM biozone divergence, previously reported as marking the timing of a major diversification among seed plants (Andrews 1963; DiMichele et al. 1989). The appearance of a differentiated sclerotesta and wing-like sarcotesta by the late Devonian represents a relatively high level of divergence compared with the variability observed among previously known, well-documented Devonian preovules. Despite the increase of morphological information on early seed plants, much uncertainty still exists in elucidating functional and life history processes. Equally difficult have been attempts to interpret possible adaptive role(s) of the key morphological structures of the seed as well as determining which selective forces were operating on the earliest seed plants (Niklas 1983a, 1985; Haig and Westoby 1989; Rothwell and Wight 1989). The ring of simple terete preintegumentary lobes observed in the earliest seed plants possibly represents a structure pre-apted for a range of possible functions following modification (Gould and Vrba 1982). Rothwell and Scheckler (1988) demonstrated evidence of developmental post-pollination shifts in preintegument arrangement around the nucellus in Elkinsia. It is unknown exactly to what extent this arrangement afforded protection to the nucellus, as the lobes are unfused for most of their length. Also this development was not observed in Moresnetia in which preintegument lobes are described as less-fused laterally. It is possible that among the earliest seed plants, preintegumentary lobes provided little or no protection to the megagametophyte and embryo. If this is the case the appearance of preintegument lobes may have been initially non-adaptive but pre-apted for a number of functions following the relevant structural modifications: fusion (protection); centripetal differentiation into wings (air mediated dispersal); development of air spaces (water mediated dispersal); or entire fusion (aerodynamic optimization for pollination). One of the most intriguing discussions concerning the early evolution of the seed habit concerns the ecological and evolutionary processes characterizing the morphogenesis of the preintegument and their consequences on pollination, protection and dispersal. Recent studies on the reproduction of early seed plants have focused on the pollination process and have determined the precise functioning of the lagenostome, pollen chamber, central column and pollen chamber floor during pre- and post-pollination phases (Rothwell and Scheckler 1988; Rothwell and Wight 1989; Rothwell and Serbet 1992). There is, however, far more uncertainty concerning the functional role of the preintegument prior to the evolution of an entire integument, whether it acted (1) as protection against desiccation, herbivory and microbial activity or (2) to maximize pollination (Niklas 1981a, 19816, 1983a, 19836, 1985) (i.e. aerodynamic optimization of the preovule for the deposition of pollen at the nucellar apex). As the individual preovule (nucellus and preintegument) represented the abscissed diaspore or unit of dispersal, a third functional role for the preintegument among early seed plants may have included one of dispersal. There is no direct evidence that any of the cupulate structures characterizing early seed plants before the late Carboniferous functioned as part of the diaspore unit and that modification of the early seed plant 590 PALAEONTOLOGY, VOLUME 40 text-fig. 4. Reconstruction of Warsteinia paprothii with prominent winged preintegument lobes ad- pressed or fused to the nucellus to just below the differentiation of the lagenostome. The inner pre- integument layer is reconstructed here as if entire around the nucellus although this is equivocal. The cut-away portion of the preintegument indicates the orientation of the linearly and perpendicularly aligned organic material of the inner and outer preintegument layers. Scale bar represents 1 mm. diaspore for dispersal would have required modification of the seed itself. Rothwell and Scheckler (1988) described attached cupules bearing only the proximal parts of seed stalks which suggests direct evidence of seed abscission/dispersal. However, they also reported isolated cupules bearing seeds and the possibility exists that cupule/seed complexes might have functioned as diaspores in Ellcinsia, or reflect a tendency towards senescence, withering and fragmentation at maturity. If the German preovules are interpreted correctly, they show a number of similarities with winged diaspore features of more recent fossils and many extant diaspores demonstrated by a fibrous network surrounded by thin membraneous material (Ridley 1930; van der Pijl 1969; Augsburger 1986; Tift'ney 1986). Recent studies have shown that the fall velocities and potential dispersal distances of extant diaspores are dependant on 'wing loading’ (expressed as the weight divided by the surface area of the diaspore) in addition to the aerodynamic movement of the diaspore during descent, which is governed primarily by the shape of the object (Green 1980; Augsburger 1986). Diaspores with similar wing-loads may differ significantly in their descent velocity on account of their variable aerodynamic movement. Numerous studies have identified the aerodynamic characteristics of modern plant diaspores, with some authors grouping similar aerodynamic movements. However, without the appropriate data concerning the relative densities and accurate three dimensional reconstruction, it is difficult to estimate what sort of aerodynamic movement these seeds would demonstrate. For example, all of the familiar autorotating diaspores examined by Augsburger (1986) possess only one wing, with the heaviest mass concentrated either in the centre (rolling autogyros) or at one end (autogyro). Neither of these broad morphologies fits that of the Devonian ROWE: DEVONIAN WINGED PREOVULES 591 preovule described here, which resembles more, in shape but not in overall size, a fruit described as a ‘tumbler' (Augsburger 1986) which have been demonstrated to descend, as implied by the name, in a more irregular fashion. The new winged seed, and others such as Lyrasperma , may show evidence of a decreased ‘wing- loading’ compared with earlier and contemporaneous non-winged preovules such as Elkinsia , Moresnetia and Archaeosperma. The development of the sarcotesta in the preintegument lobes might significantly decrease the mass-to-area ratio, provided that the mass of the nucellus in Warsteinia was equivalent to that of other non-winged forms. As well as decreasing fall velocity, a winged morphology as shown could alternatively function by tumbling or saltation along the ground propelled by air movements (Ridley 1930) but the relative significance of either proposed dispersal optimizations is, of course, speculative. The reduction in fall velocity might be insignificant if the parent plant was of limited height of only one to several metres, as is presumed for slender early seed plants. The difficulty in assessing the real performance of winged diaspore structures among living forms, let alone fossils, is aptly demonstrated in studies of fruit dispersal in Acer saccharinum , in which it has been found that the preferential timing of abscission at higher rather than ambient wind velocities effects the release of disapores, and that the drag created by the attached winged fruit is an important factor in this mechanism (Greene and Johnson 1992). The presence of winged seed diaspore structures by the late Devonian could be interpreted as an amelioration of a less derived preovule morphology with the possibility of, in one way or another, optimizing dispersal whether by influencing fall velocity, ground saltation, drag-related abscission or a combination of these. In terms of size, the preovules are consistent with the other contemporaneous Devonian and many Lower Carboniferous preovules. In a discussion of seed dispersal in the fossil record, Tiffney (1986) demonstrated that estimated volumes of Devonian preovules were approximately between 1 and 10 mm3. This range would include the nucellar region of Warsteinia if reconstructed as an ovoid three-dimensional structure. Between the earliest records of seed plant preovules in the upper Famennian, characterized by open preintegument lobes, and the late Carboniferous trend towards an entire integument which has characterized all seed plants since, a wide variability of preovule preintegument forms existed among seed plants retaining hydrasperman reproduction. Attempts to explain the adaptive significance of the preintegument should focus on (1) the evolutionary appearance of a lobed preintegument and (2) the morphological diversification following the establishment of preovule organization in seed plants. In terms of an adaptive interpretation of the appearance of the preintegument, a general consensus appears to indicate no strong evidence for either a significantly protective role, which may have been carried out adequately by the megasporangial wall, as in other heterosporous groups, or for optimizing pollen capture. Either or both of these processes may have played important roles later during the diversification of hydrasperman seed plants prior to entire integumentation. A selective pressure for optimizing food reserve availability has also been attributed to the preintegument in partitioning available food reserves and favouring the provisioning of pollinated/fertilized rather than unfertilized preovules (Westoby and Rice 1982; Haig and Westoby 1989). It is uncertain whether such a mechanism would have been functional among the earliest seed plants in which tissue differentiation of the preintegument lobes is simple and intimately associated with the nucellus only in the basal area. As with other possible functions of the preintegument, this could have been significant in more derived seed plant organizations where the preintegument was intimately fused with the nucellar tissue. The interpretation proposed here is that the appearance of terete preintegument lobes could have been initially non-adaptive, but open to a range of possible mutational and developmental shifts in terms of fusion, tissue proliferation and changes in the amount and positioning of the vascular supply. The de novo evolution of the preintegument radically enlarged the potential for elaboration and fine-tuning of life history processes. The adaptive or non-adaptive nature of other ‘key’ elements of the seed habit include processes concerned with the apical opening of the nucellus and the lagenostome, pollen chamber, pollen chamber floor and central column mechanism attributed to plants with hydrasperman reproduction (Rothwell 1986). The evolution of the pollination 592 PALAEONTOLOGY, VOLUME 40 mechanism allowed early seed plants to jump the ‘adaptive valley’ separating water dependant, free-sporing heterospory from the seed habit (Chaloner and Pettitt 1987; Bateman and DiMichele 1994). Whether all the elements of the hydrasperman apparatus listed above appeared simultaneously or whether another configuration such as that observed in Coumiasperma remyii may have preceded it remains to be elucidated. Unlike the difficulty in assigning an unequivocal function to the preintegument, the pollination mechanism represented by a simple distal aperture is more easily interpreted as being spontaneously adaptive, even if initially less complex than the full suite of characters comprising hydrasperman reproduction outlined above. The range of preintegument variations among hydrasperman seed plants prior to the development of the entire integument and integumental micropyle is striking, but, like the difficulties in interpreting the original appearance of the preintegument, the ecological and selective scenarios explaining this diversity have recently been hotly debated. This uncertainty is basically polarized into two general themes. Firstly, that the range of forms are representative of selective conditions characterized by a lack of biotic competition : an open selective landscape with a range of vacant niches (DiMichele et al. 1989). Another interpretation is that the range of variability observed among preovule types represents specialized adaptations fitted for specific niches and consequently consistent with conditions of biotic competition (e.g. Andrews 1963; Rothwell and Scheckler 1988). Because of the shortage of palaeoecological information and whole plant reconstructions from many of these early seed plant assemblages, the range of niches available and/or exploited and the nature of any biotic and abiotic stresses are largely unknown. Furthermore, the morphology and organization of the preovule is, of course, only one of many characteristics contributing to that plant’s overall fitness. Evidence from other studies indicates that by the late Tournaisian, hydrasperman seed plants probably comprised a relatively wide range of growth habits and physical niches from large self-supporting, arboreous trees, with small fronds and leaf abscission, to slender semi-self-supporting plants bearing large fronds (Gordon 1935; Long 1979; Galtier 1992; Rowe et al. 1993; Speck and Rowe 1994). Provisional reconstructions of several of these plants (Retallack and Dilcher 1988) indicate that all probably possessed hydrasperman preovules differing in preintegument and cupule morphology, but that all possessed the same type of hydrasperman pollination mechanism. It would be surprising, given the range of these growth forms by the end of the Tournaisian, if there was no biotic competition for niches among plants with a seed habit. There is no reason also why the increase in biotic competition characterizing the latter part of this scenario should be only between early pteridosperms and not between pteridosperms and other contemporaneous elements such as rhacophytopsid ferns, as there is some palaeoecological evidence that the latter were in close succession ecologically and temporally to stands of early seed plants (Scheckler 1986; Rothwell and Scheckler 1988). Another possible selective scenario is that the absence of a dormancy mechanism (Tiffney 1986; Mapes et al. 1989) and germination to specific environmental cues might have been an important constraint on the direction of life history strategies among early seed plants. If embryo development occurred continuously after fertilization, presumably while still attached to the parent sporophyte, there would have been a trade-off between the advantages of size increase and ‘rigour’ of the embryo against the increasing mass of the potential diaspore and its corresponding decrease in dispersal potential. The general absence of embryos preserved within preintegumented hydra- sperman diaspores in the fossil record and the abundance of preserved pre-pollination stages has been discussed as evidence of an absence of dormancy among early hydraspermans prior to the earliest evidence of arrested development in a seed plant embryo in a Late Carboniferous cordaitean seed (Tiffney, 1986; Mapes et al. 1989). Coupled with unarrested embryo development is the dispersal potential of the earliest hydrasperman seeds, and the effects of mortality factors connected with density and distance in relation to the maternal plant or between siblings with regard to shading, pathogens, predators and competitors. Besides this, the absence of a dormancy mechanism and control of germination, during arguably the most vulnerable point in the seed plant life cycle, may indicate that the potential of the earliest hydrasperman seed plants to invade drier habitats might be overrated and that their reproductive success lies elsewhere, such as in the sheer output ROWE: DEVONIAN WINGED PREOVULES 593 of successfully fertilized diaspores. It is arguable whether the early stages of development of the hydrasperman seedling emerging from the preintegumented seed would have had more advantage in terms of size and presumed storage capability of metabolic reserves over early sporophyte growth stages implanted on exosporic gametophytes. Coupled with the constraints on development and dispersal outlined above, and linked with absence of dormancy, the size and likely nutrient storage properties of the earliest seeds could therefore be interpreted as another adaptive ‘valley’ in the transition towards a full seed habit superimposed on that concerned with pollen transfer and megasporangium/preovule shape. Establishment and emergence without a control of germination and dormancy, would represent an additional suite of constraints and possible adaptations among the earliest seed plants. The generally small size of the earliest seeds may therefore give some credence to the hypothesis of selective pressure for increased nutrient storage (Haig and Westoby 1989) possibly resulting from germination and development without dormancy control. Acknowledgements. This work was carried out at the Lehr- und Forschungsgebiet fur Geologie und Palaontologie, RWTH, Aachen during the tenure of a Humboldt Foundation Fellowship and hosted by G. Flajs to whom I am most grateful. I also wish to acknowledge the kind assistance of E. Paproth and D. Korn who introduced me to the area and the site, assistance in the field by J. Weber, and preparation of the diagrams by L. Meslin. I thank G. Rothwell and C. Cleal for constructive comments on the manuscript. The work was completed at the Dept of Geology, Royal Holloway and Bedford New College, Egham, UK during a NERC postdoctorate fellowship, hosted by A. C. Scott, which is gratefully acknowledged. This is contribution No. 96- 074, ISEM, UMR5554, CNRS, Montpellier. REFERENCES Andrews, H. N. 1963. Early seed plants. Science , 142, 925-931. arber, e. a. n. and goode, r. h. 1915. On some fossil plants from the Devonian rocks of North Devon. Proceedings of the Cambridge Philosophical Society , 18, 89 104. Arthur, w. 1984. Mechanisms of morphological evolution: a combined genetic , developmental and ecological approach. John Wiley, New York, 275 pp. 1988. A theory of the evolution of development . John Wiley, New York, 94 pp. augsburger, c. K. 1986. Morphology and dispersal potential of wind-dispersed diaspores of neotropical trees. American Journal of Botany. 73, 353-363. Barnard, p. d. w. and long, a. g. 1975. Triradioxylon - a new genus of Lower Carboniferous petrified stems and petioles together with a review of the classification of early pterophytina. Transactions of the Royal Society of Edinburgh , 69, 231-250. bateman, r. m. and dimichele, w. a 1994. Heterospory: the most iterative key innovation in the evolutionary history of the plant kingdom. Biological Review , 69, 345-417. calder, m. g. 1938. On some undescribed species from the Lower Carboniferous flora of Berwickshire; together with a note on the genus Stenomyelon Kidston. Transactions of the Royal Society of Edinburgh. 59, 309-331 chaloner, w. g., hill, a. j. and lacey, w. s. 1977. First Devonian platyspermic seed and its implications in gymnosperm evolution. Nature. 265, 233-235. and pettitt, j. M. 1987. The inevitable seed. Bulletin de la Societe Botanique de France , 134, Actualites Botaniques 1987. 2. 39-49. dimichele, w. a., davis, J. i. and olmstead, r. g. 1989. Origins of heterospory and the seed habit: the role of heterochrony. Taxon. 38, 1-1 1 fairon-demaret, m. and scheckler, s. E. 1987. Typification and redescription of Moresnetia zalesskyi Stockmans, 1948, an early seed plant from the Upper Famennian of Belgium. Bulletin de I'Institut des Sciences Naturelles de Belgique. Sciences de la Terre. 57, 183-199. galtier, j. 1992. On the earliest arborescent gymnosperms. Courier Forschungsinstitut Senckenberg. 147, 119-125. and Rowe, n. p. 1989. A primitive seed-like structure and its implications for early gymnosperm evolution. Nature. 6230. 225-227. 1991. A new permineralized seed-like structure from the basalmost Carboniferous of France. Neues Jahrbuch fur Geologie und Palaontologie, Abliandlungen, 183, 103 120. 594 PALAEONTOLOGY, VOLUME 40 Gillespie, w. h., rothwell, G. w. and scheckler, s. E. 1981. The earliest seeds. Nature , 293, 462^464. Gordon, w. r. 1935. The genus Pitys, Whitham, emend. Transactions of the Royal Society of Edinburgh , 58, 279-311. gould, s. j. and vrba, e. s. 1982. Exaptation - a missing term in the science of form. Palaeobiology, 8, 4-15. green, d. s. 1980. The terminal velocity and dispersal of spinning samaras. American Journal of Botany , 67, 1218-1224. greene, d. f. and Johnson, e. a. 1992. Fruit abscission in Acer saccharinum with reference to seed dispersal. Canadian Journal of Botany, 70, 2277-2283. haig, d. and westoby, m 1989. Selective forces in the emergence of the seed habit. Biological Journal of the Linnean Society , 38, 215-238. higgs, k. and streel, m. 1984. Spore stratigraphy at the Devonian-Carboniferous boundary in the northern ‘Rheimsches Schiefergebirge’, Germany. Courier Forschungsinstitut Senckenberg , 67, 157-179. keupp, h. and kompa, r. 1984. Mikrofazielle und Sedimentologische Untersuchungen an Devon/Karbon Profilen am Nordrand des Rechtsrheinischen Schiefergebirges. Courier Forschungsinstitut Senckenberg , 67, 139-142. long, a. G. 1959. On the structure of Calymmatotheca kidstoni Calder (emended) and Genomosperma latens gen. et sp. nov. from the Calciferous Sandstone Series of Berwickshire. Transactions of the Royal Society of Edinburgh , 64, 29-44. 1960a. On the structure of Samaropsis scotica Calder (emended) and Euryostoma angulare gen. et sp. nov., petrified seeds from the Calciferous Sandstone Series of Berwickshire. Transactions of the Royal Society of Edinburgh , 64, 261-280. - 19606. Stamnostoma huttonense gen. et sp. nov. - a pteridosperm seed and cupule from the Calciferous Sandstone Series of Berwickshire. Transactions of the Royal Society of Edinburgh, 64, 201-215. 1961. Some pteridosperm seeds from the Calciferous Sandstone Series of Berwickshire. Transactions of the Royal Society of Edinburgh , 64, 401-419. 1969. Euryostoma trigona sp. nov., a pteridosperm ovule borne on a frond of Alcicornopteris Kidston. Transactions of the Royal Society of Edinburgh , 68. 171-182. 1975. Further observations on some Lower Carboniferous seeds and cupules. Transactions of the Royal Society of Edinburgh , 69, 267-293. \911a. Some Lower Carboniferous pteridosperm cupules bearing ovules and microsporangia. Transactions of the Royal Society of Edinburgh, 70, 1—11. 19776. Lower Carboniferous pteridosperm cupules and the origin of angiosperms. Transactions of the Royal Society of Edinburgh, 70, 13-35. 1979. Observations on the Lower Carboniferous genus Pitus Witham. Transactions of the Royal Society of Edinburgh, 70, 111-127. luppold, F. w., clausen, c. d., korn, d. and stoppel, d. 1994. Devon/Karbon-Grenzprohle im Bereich von Remscheid-Altenaer Sattel, Warsteiner Sattel, Briloner Sattel und Attendorn-Elsper Doppelmulde (Rheinisches Schiefergebirge). 7-70. In korn, d., clausen, c. d. and luppold, f. w. (eds). Die Devon / Karbon-Grenze im Rheinschen Schiefergebirge. Landschaftsverband Westfalen-Lippe, Munster, 221 pp. mapes, G., rothwell, G. w. and haworth, m. T. 1989. Evolution of seed dormancy. Nature, 337, 645-646. matten, l. c. and lacey, w. s. 1981. Cupule organization in early seed plants. 221-234. In romans, r. c. (ed.). Geobotany II. Plenum, New York. and lucas, r. c. 1980a. Studies on the cupulate genus Hydrasperma Long from Berwickshire and East Lothian in Scotland and County Kerry in Ireland. Botanical Journal of the Linnean Society, 81, 249-273. may, b. i. and lucas, R. c. 19806. A megafossil flora from the Uppermost Devonian near Ballyheigue, Co. Kerry, Ireland. Review of Palaeobotany and Palynologv , 29, 241-251. tanner, w. r. and lacey, w. s. 1984. Additions to the silicified Upper Devonian/Lower Carboniferous flora from Ballyheigue, Ireland. Review of Palaeobotany and Palynologv, 43, 303-320. niklas, K. 1981a. Simulated wind pollination and airflow around ovules of some early seed plants. Science, 211, 275-277. 19816. Airflow patterns around some early seed plant ovules and cupules: implications concerning efficiency in wind pollination. American Journal of Botany, 68, 635-650. 1983a. Early seed plant wind pollination studies: a reply. Taxon, 32, 99-100. 19836. The influence of Paleozoic ovule and cupule morphologies on wind pollination. Evolution, 37, 968-986. ROWE: DEVONIAN WINGED PREOVULES 595 1985. The aerodynamics of wind pollination. The Botanical Review , 51, 328-386. — 1992. Plant biomechanics : an engineering approach to plant form and function. University of Chicago Press, Chicago, 607 pp. pettitt, j. m. and beck, c. b. 1968. Archaeosperma arnoldii - a cupulate seed from the Upper Devonian of North America. Contributions to the Museum of Paleontology, University of Michigan , Ann Arbor , 22, 139-154. pul, L. van der 1969. Principles of dispersed in higher plants. Springer Verlag, Berlin, 153 pp. retallack, G. J. and dilcher, d. l. 1988. Reconstructions of selected seed ferns. Annals of the Missouri Botanical Gardens , 75, 1010-1057. ridley, H. N. 1930. The dispersal of plants throughout the world. L. Reeve and Co., Ashford, 744 pp. Rogers, i. 1926. On the discovery of fossil fishes and plants in the Devonian rocks of North Devon. Transactions of the Devonshire Association for the Advancement of Science, Literature and Art , 58, 223-234. rothwell, G. w. 1986 Classifying the earliest gymnosperms. 137-161. In spicer, r. a. and thomas, b. a. (eds). Systematic and taxonomic approaches in palaeobotany. Oxford University Press, Oxford, 321 pp. and scheckler, s. e. 1988. Biology of ancestral gymnosperms. 85-134. In beck, c. b. (ed. ). Origin and evolution of gymnosperms. Columbia University Press, New York, 504 pp. — and gillespie, w. h. 1989. Elkinsia gen nov., a Late Devonian gymnosperm with cupulate ovules. Botanical Gazette, 150. 170-189. - and serbet, r. 1992. Pollination biology of Elkinsia polymorpha , implications for the origin of gymnosperms. Courier Forschungsinstitut Senckenberg, 147, 225-231. 1994. Lignophyte phylogeny and the evolution of spermatophytes: a numerical cladistic analysis. Systematic Botany, 19, 443-482. — and wight, d. c. 1989. Pullaritheca longii gen. nov. and Kerryia mattenii gen. et sp. nov., Lower Carboniferous cupules with ovules of the Hvdrasperma tenuis-type. Review of Palaeobotany and Palynologv, 60, 295-309. rowe, n. p. 1 992c/. Winged Late Devonian seeds. Nature, 359, 682. - 1992A Permineralized Tournaisian plants from Oberrodinghausen in the ‘Rheinisches Schiefergebirge’, Germany. Courier Forschungsinstitut Senckenberg, 147. 127-135. and galtier, j. 1990. A Lower Carboniferous plant assemblage from La Serre (Montagne Noire, France). Part II. Gymnosperms. Review of Palaeobotany and Palynologv, 63, 91-1 15. speck, t. and galtier, j. 1993. Biomechanical analysis of a Palaeozoic gymnosperm stem. Proceedings of the Royal Society of London, 252, 19-28. scheckler, s. e. 1986. Geology, lloristics and paleoecology of Late Devonian coal swamps from Appalachian Laurentia (U.S.A.). Annates de la Societe Geologique Belgique, 109, 209-236. speck, t. and rowe, n. p. 1994 Biomechanical analysis of Pitus davi: early seed plant vegetative morphology and its implications on growth habit. Journal of Plant Research, 107, 443^160. stewart, w. n. 1983. Paleobotany and the evolution of plants. Cambridge University Press, Cambridge, 405 pp. stockmans, f. 1948. Vegetaux du Devonien superieur de la Belgique. Memoires du Musee Roved if Histoire Naturelle de Belgique, 110, 1-85. tiffney, b. h. 1986. Evolution of seed dispersal syndromes according to the fossil record. 273-305. In Murray, d. r. (ed.). Seed dispersal. Academic Press, Sydney, Australia. 322 pp. westoby, m. and rice, b. 1982. Evolution of seed plants and inclusive fitness of plant tissues. Evolution, 36, 713-724. N. P. ROWE Laboratoire de Paleobotanique Institut des Sciences de l’Evolution (UMR 5554 CNRS) Place E. Bataillon, CP 062 Universite Montpellier 2 Typescript received 20 February 1996 34095 Montpellier Cedex 05 Revised typescript received 19 August 1996 France BIBLIOGRAPHY AND INDEX OF CATALOGUES OF TYPE, FIGURED AND CITED FOSSILS IN MUSEUMS IN GREAT BRITAIN AND IRELAND (SUPPLEMENT 1975-1996) by VALERIE K. DEISLER and MICHAEL G. BASSETT Abstract. Data on the substantial holdings of type, figured and cited fossils in many institutions in Great Britain and Ireland are summarized in numerous published catalogues, which are an indispensable aid in tracing material especially for taxonomic studies. Taxonomic, stratigraphical and museum location indexes are provided for catalogues published over the past 20 years. Some useful unpublished reports containing similar information are also noted. A previous collation of documented holdings of type, figured and cited fossils in museums and other institutions throughout the British Isles listed over 100 catalogues produced since the early 1890s (Bassett 1975). Over the past 20 years or so this data base has expanded considerably with the publication of about three-quarters as many again catalogues, notes and reports that identify collections or individual specimens within these categories. Reasons for this significant expansion are varied, but they are welcome in demonstrating the growth of collective institutional and individual responsibilities to ensure that essential reference collections are properly curated and conserved for the future, and that details of them are widely publicized (see Bassett 1975, pp. 754-755). The compilations continue to form indispensable reference sources for palaeontologists. In addition, they also serve to emphasize the growing acceptance of responsibility by many institutions in meeting the requirements for safeguarding type material as summarized under Recommendation 72G of the International Code of Zoological Nomenclature (International Trust for Zoological Nomenclature 1985, p. 147). It is relevant here to repeat this Recommendation of the Code which states that: Every institution in which name-bearing types are deposited should (1) ensure that all are clearly marked so that they will be unmistakably recognized as name-bearing types; (2) take all necessary steps for their safe preservation; (3) make them accessible for study; (4) publish lists of name-bearing types in its possession or custody; and (5) so far as possible, communicate information concerning name-bearing types when requested. Much of the impetus for the growing production of these data in the British Isles stems from the work and encouragement of the Geological Curators’ Group (GCG), formed in 1974 and working since 1976 as a Specialist Group of the Geological Society, London. Initially through its Newsletter , and since 1980 via its more formal publication. The Geological Curator , the GCG has regularly published catalogue data and tracked down many individual specimens and collections that were previously ‘missing’ or presumed to be irretrievably lost, including material in private collections. Among the positive results of these detailed searches is the fact that a good many specimens in private collections have now been donated to institutions for safe keeping and wider availability for scientific study, and some organizations without properly trained palaeontological [Palaeontology, Vol. 40, Part 2, 1997, pp. 597— 617[ © The Palaeontological Association 598 PALAEONTOLOGY, VOLUME 40 and/or conservation staff have similarly transferred material to museums in order to ensure the same aims. The influential survey and report on the state and status of geological collections throughout UK museums, also generated by the GCG (Doughty 1981), gave added stimulus to museums to investigate and to document properly their holdings of fossils. This report discovered, for example, that of 64 museums known to house type fossil material, 35 had no qualified curatorial staff, and many others of the 288 admitting to housing geological collections knew very little in detail of their contents. Activities of GCG members were focused sharply by the Doughty report in order to remedy these deficiencies. At the same time, activities of the larger museums with full time curatorial staff continued to include the publication of type/figured/cited catalogues. Collectively, this multitude of activity since 1975 forms the basis for this bibliography and index, although a few pre- 1975 publications overlooked in the earlier compilation (Bassett 1975), are also included. BIBLIOGRAPHY Most of the publications listed here contain inventory data for individual fossil specimens with information on the repository, museum registration numbers, type data (where applicable), and details of page, plate, and figure numbers in a previous publication referring to the particular specimen. Descriptive taxonomic publications are not included. In the few cases where all the inventory/reference data are not quoted, details will generally allow an individual specimen to be identified. adams, c. G. 1960. A note on two important collections of Foraminifera in the British Museum (Natural History). Micropaleontology , 6, 417^418. - harrison, c. a. and hodgkinson, r. L. 1980. Some primary type specimens of Foraminifera in the British Museum (Natural History). Micropaleontology , 26, 1-16. Andrews, s. m. 1982. 38-51. In The discovery of fossil fishes in Scotland up to 1845 with checklists of Agassiz's figured specimens. Royal Scottish Museum Studies, Royal Scottish Museum, Edinburgh, 87 pp. Baird, w. j. 1980. A catalogue of trilobites in the Royal Scottish Museum, Edinburgh. Royal Scottish Museum Information Series, Geology, 8, i-iv, 1-72. benton, m. j. 1979. H. A. Nicholson (1844-1899), invertebrate palaeontologist: bibliography and catalogue of his type and figured material. Royal Scottish Museum Information Series, Geology, 1, i-viii, 1-94. - and trewin, n. h. 1978. Catalogue of the type and figured material in the Palaeontology Collection, University of Aberdeen, with notes on the H. A. Nicholson collection. Publications of the Department of Geology and Mineralogy, University of Aberdeen, 2, i-iv, 1-28. boyd, m. J. 1983. Catalogue of type, figured and cited fossils in Kingston upon Hull City Museums. Geological Curator, 3, 476 — 485. [ + 3 pp. undated typescript Supplements Nos 1-3]. - 1986. Supplement to the catalogue of the Carboniferous amphibians in the Hancock Museum, Newcastle upon Tyne. Transactions of the Natural History Society of Northumbria, 46 (Supplement), 1-8. - and turner, s. 1980. Catalogue of the Carboniferous amphibians in the Hancock Museum, Newcastle upon Tyne. Transactions of the Natural History Society of Northumbria, 46, 1-24. butler, d. 1980. Collections and collectors of note. 38. North Devon Athenaeum, Barnstaple. B) Figured Devonian fossils in the collections. Geological Curator, 2, 588-592. Campbell, e. 1976. Catalogue of type and figured fossils in Glasgow Museum. 59 pp. [unpublished typescript catalogue; limited distribution]. chandler, g. and Hannah, i. c. 1949. [List of type and figured specimens in the Dudley collection]. 5-9. In Dudley as it was and as it is today. Batsford, London, xii + 208 pp. Clark, r. d 1982. Type, figured and cited Jurassic Cephalopoda in the collection of the Institute of Geological Sciences. Report of the Institute of Geological Sciences, 82/9, ii + 104 pp. cocks, l. r. m. 1978. A review of British Lower Palaeozoic brachiopods, including a synoptic revision of Davidson’s Monograph. Monograph of the Palaeontographical Society, 131 (549), 1-256. crane, m. d. 1980u. An annotated list of material in the City of Bristol Museum and Art Gallery collected by T. R. Fry. Geological Curator, 2, 563-571. - 19806. Catalogue of type, figured and cited fossils in the City of Bristol Museum and Art Gallery. Part DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 599 1, Plantae. Geological Curator , 2, Supplement, 1-17, i-iv. — and getty, t. a. 1975. Geological collections and collectors of note. 8. An historical account of the palaeontological collections formed by R. W. Hooley (1865-1923). Newsletter of the Geological Curators' Group, 4. 170-179. CRAWLEY, M. 1988. Catalogue of the type and figured specimens of macrofossil algae in the British Museum (Natural History). British Museum (Natural History), London, 52 pp. cross, t. 1975a. Type, figured and cited material in the palaeontological collections of the City Museum, Peterborough. Newsletter of the Geological Curators' Group, 4, 180-183. — 19756. A catalogue of the fossil vertebrates in the City Museum , Peterborough. Part One , Reptiles and Fish. City Museum and Art Gallery, Peterborough, 21 pp. duffin, c. 1978. Collections and collectors of note. 4. The Bath geological collections. The importance of certain vertebrate fossils collected by Charles Moore: an attempt at scientific perspective. Newsletter of the Geological Curators' Group, 2, 59-67 . — 1979. Collections and collectors of note. 4. The Bath geological collections. The Moore collection of Upper Liassic crocodiles: a history. Newsletter of the Geological Curators' Group, 2, 235-252. eagar, m. and preece, r. 1977. Collections and collectors of note. 14. The Manchester Museum. B. List of type specimens in the Museum since 1952. Newsletter of the Geological Curators' Group, 11, 25-33. ensom, p. c. 1983. Lost and found. 140. Silvester, N. L. Geological Curator, 3, 489. garrad, l. s. 1979. Collections and collectors of note. 16. The Manx Museum shells from ‘The Manxland Drift’. Newsletter of the Geological Curators' Group, 2, 231—232. Hancock, E. G., howell, a. and torrens, H. s. 1976. Geological collections and collectors of note: I I . Bolton Museum : 3. Palaeontological type material so far recognised in Bolton Museum. Newsletter of the Geological Curators' Group, 7, 332-335. hodgkinson, r. l. 1995. The Hull University collection of Ostracoda in the Natural History Museum, London; sources of type and figured material. Micropaleontology , 41, 381-382. joysey, K. a. 1960. Note on the Brady Collection of Foraminifera. Micropaleontology, 6, 416. knell, s. 1986. Abingdon’s Arkell ammonites. Geological Curator, 4, 510. lewis, D. N. 1986. Catalogue of the type and figured specimens of fossil Echinoidea in the British Museum ( Natural History). British Museum (Natural History), London, 85 [+10] pp., 5 pis. — 1993. Catalogue of the type and figured specimens of fossil Asteroidea and Ophiuroidea in the Natural History Museum. Bulletin of the Natural History Museum, Geology Series, 49, 47-80. loeffler, e. j. and crane, m. d. 1982. Catalogue of type, figured and cited fossils in the City of Bristol Museum and Art Gallery. Part 2, Invertebrata : Porifera, Coelenterata, Bryozoa. Geological Curator , 3, Supplement, 19-37. mancenido, m. o. and damborenea, s. e. 1978. Comments on some type and figured brachiopods and bivalves in the Yorkshire Museum. Newsletter of the Geological Curators' Group, 2, 122-123. mitchell, m. 1986. The fossil collection of C. B. Salter, from Cliff Quarry, Compton Martin, Mendip Hills. Geological Curator, 4, 487-491. morris, s. f. 1980. Catalogue of the type and figured specimens of fossil Crustacea ( excluding Ostracoda), Chelicerata, Myriapoda and Pycnogonida in the British Museum (Natural History). British Museum (Natural History), London, iv + 53 pp. — 1988. A review of British trilobites including a synoptic revision of Salter’s monograph. Monograph of the Palaeontographical Society, 140(574), 1-316. — and fortey, r. a. 1985. Catalogue of the type and figured specimens of Trilobita in the British Museum (Natural History). British Museum (Natural History), London, 183 pp., 8 pis. Murray, J. w. and taplin, c. m. 1984. The W. B. Carpenter Collection of Foraminifera: a catalogue. Journal of Micropalaeontology, 3, 55-58. newman, A. and chatt-ramsey, J. 1988. A catalogue of the specimens figured in ' The Fossil Flora' by John Lindley ( 1799-1865 ) and William Hutton (1799-1860) held by the Hancock Museum, Newcastle upon Tyne , including a biography of William Hutton. The Hancock Museum, Newcastle upon Tyne, viii + 67 pp. nudds, j. r. 1982a. Catalogue of type, figured and referred fossils in the Geological Museum of Trinity College, Dublin: Part 1 (Protozoa, Porifera, Archaeocyatha, Coelenterata, Bryozoa). Journal of Earth Sciences , Royal Dublin Society, 4. 133-165. — 19826. Catalogue of type, figured and referred fossils in the Geological Museum of Trinity College, Dublin: Part 2 (Brachiopoda, Mollusca). Journal of Earth Sciences, Royal Dublin Society, 5, 61-89. — 1982c. Catalogue of type and figured corals from the Geological Museum, Trinity College, Dublin. Fossil Cnidaria, 11, 19-26. 600 PALAEONTOLOGY, VOLUME 40 — 1983. Catalogue of type, figured and referred fossils in the Geological Museum of Trinity College, Dublin : Part 3 (Arthropoda, Echinodermata, Graptoloidea, Conodontophorida, scolecodonts, Chitinozoa, Problematica, symbiosis, trace fossils, Vertebrata). Journal of Earth Sciences, Royal Dublin Society, 5, 153-190. — 1984. Catalogue of type, figured and referred fossils in the Geological Museum of Trinity College, Dublin: Part 4 (Plantae). Irish Journal of Earth Sciences, 6, 47-93. [ Note: Nudds 1982a, 1 982A, 1983, 1984 were also re-issued, without change of pagination, by Trinity College, Dublin, as a single bound volume: Catalogue of type, figured and referred fossils in the Geological Museum of Trinity College, Dublin}. — 1988. Catalogue of type, figured, and referred fossils in the Geological Museum of Trinity College, Dublin: Supplement (Animalia). Irish Journal of Earth Sciences, 9, 177-196. — 1989. Catalogue of type, figured, and referred fossils in the Geological Museum of Trinity College, Dublin: Supplement (Plantae). Irish Journal of Earth Sciences, 10, 43-53. — 1992a. Catalogue of type, figured and referred fossils in the Geological Department of the Manchester Museum. Proceedings of the Yorkshire Geological Society, 49, 81-94. — 1992 b. The R. M. C. Eagar Collection of non-marine bivalves; type and figured specimens in the Geological Department of the Manchester Museum. Manchester Museum Publications, New Series, No. NS. 6. 92. [microfiche]. OWENS, R. M. and BASSETT, m. g. 1995. Catalogue of type, figured and cited fossils in the National Museum of Wales. Supplement 1971-1994. National Museum of Wales, Geological Series No. 12, Cardiff, 250 pp. parkes, M. a. and sleeman, a. G. 1997. Catalogue of type, figured and cited fossils in the Geological Survey of Ireland. Geological Survey of Ireland. paton, r. L. 1976. A catalogue of fossil vertebrates in the Royal Scottish Museum, Edinburgh. Part Five. Acanthodii. Royal Scottish Museum Information Series, Geology, 6, i-viii, 1-40. — 1981. A catalogue of fossil vertebrates in the Royal Scottish Museum, Edinburgh. Part Six. Placodermi. Royal Scottish Museum Information Series, Geology, 9, i-viii, 1-70. pattison, J. 1977. Catalogue of the type, figured and cited specimens in the King Collection of Permian fossils. Bulletin of the Geological Survey of Great Britain , 62, 33 — 44. Phillips, D. 1977. Catalogue of the type and figured specimens of Mesozoic Ammonoidea in the British Museum ( Natural History). British Museum (Natural History), London, iv + 220 pp. — 1982 a. Catalogue of the type and figured specimens of fossil Cephalopoda ( excluding Mesozoic Ammonoidea ) in the British Museum ( Natural History). British Museum (Natural History), London, 94 pp. — 19826. Additions to the catalogues of type and figured fossil Cephalopoda in the British Museum ( Natural History). British Museum (Natural History), London, 155 pp. Phillips, p. w. 1976. Geological collections and collectors of note: 10. Merseyside County Museums: B. List of type, figured and cited fossils. Newsletter of the Geological Curators' Group, 6, 270-286. powell, H. p. and edmonds, j. m. 1978. List of type-fossils in the Philpot Collection, Oxford University Museum. Proceedings of the Dorset Natural History and Archaeological Society, 98, 48-53. pyrah, b. J. 1976. Catalogue of type and figured fossils in the Yorkshire Museum: Part 1, Porifera, Coelenterata, Bryozoa, Annelida, Brachiopoda, Crustacea, Insecta. Proceedings of the Yorkshire Geological Society, 41, 35 — 47. — 1977. Catalogue of type and figured fossils in the Yorkshire Museum: Part 2, Echinodermata, Bivalvia. Proceedings of the Yorkshire Geological Society, 41, 241-260. — 1978. Catalogue of type and figured fossils in the Yorkshire Museum: Part 3, Gastropoda, Polyplacophora, Scaphopoda, Cephalopoda. Proceedings of the Yorkshire Geological Society, 41, 437 460. — 1979a. Catalogue of type and figured fossils in the Yorkshire Museum: Part 4, Pisces, Reptilia, Aves, Mammalia, Plantae. Proceedings of the Yorkshire Geological Society, 42, 4 1 5 — 437. — 19796. Lingula parallela Phillips. Type material in the Yorkshire Museum: a reply. Newsletter of the Geological Curators ' Group, 2, 184-185. radley, J. d. 1996. Type, figured and cited specimens in the Museum of Isle of Wight Geology (Isle of Wight, England). Geological Curator, 6, 187-193. riley, T. h. 1975. Geological and other collections of Henry Clifton Sorby. Newsletter of the Geological Curators' Group, 3, 130-131. rolfe, w. d. i„ ingham, j. k., currie, E. D., Neville, s., brannan, J. and Campbell, e. 1981. Type specimens of fossils from The Hunterian Museum and Glasgow Art Gallery and Museum. Hunterian Museum, University of Glasgow, 8 pp. + 5 microfiches. [Glasgow’s catalogue is also available in typescript version]. DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 601 sarjeant, w. A. s. 1983. British fossil footprints in the collections of some principal British Museums. Geological Curator , 3, 541-560. smith, J. D. d. 1989. The Silurian System by Roderick Impey Murchison. A catalogue of the fossils illustrated in Part II. British Geological Survey Research Report, SH/89/1, Stratigraphy Series, i-x, 1-211. — 1996. The Silurian System by Roderick Impey Murchison. A catalogue of the fossils illustrated in Part II. Amendments and Additions. Supplement to British Geological Survey Research Report, SH/89/1, Stratigraphy Series, 9 pp. [Unpublished typescript catalogue, Feb. 1996]. steward, d. i. and torrens, h. s. 1985. Lost and found. 47. F. Holt. Geological Curator, 4, 343. strachan, i. 1979. Collections and collectors of note. 25. Birmingham University Geological Museum. Newsletter of the Geological Curators' Group , 2, 309-321. Sutherland, a. G. 1991. A catalogue of Carboniferous corals in the National Museums of Scotland. (Based on an original catalogue by I. F. Sime, 1972). National Museums of Scotland Information Series, 9, 1-46. torrens, h. s. 1978 a. Collections and information found. 52. Holland, Harriet Sophia (c. 1835-1908) later Mrs Hutton mother of 52a. Hutton, Harriet Mary (1873-1937). Newsletter of the Geological Curators' Group , 2, 128-129. — 19786. The Sherborne School Museum and the early collections and publications of the Dorset Natural History and Antiquarian Field Club. Proceedings of the Dorset Natural History and Archaeological Society, 98, 32-42. — 1979. Collections and information found. 79. Capewell, L. P. of Dudley. Newsletter of the Geological Curators' Group , 2, 355. — 1983. Collections and information found. 141. Wellcome Institute Geological Collection. Geological Curator , 3, 494. — and taylor, M. a. 1988. Collections, collectors and museums of note. No. 55. Geological collectors and museums in Cheltenham 1810-1988. A case history and its lessons. Geological Curator, 5, 175-213. TRIPP, R. p. and HOWELLS, Y. 1981 . Catalogue of the described, figured and cited Ordovician and Silurian trilobites from the Girvan district, Scotland in the British Museum ( Natural History ). British Museum (Natural History), London, 6 foldover pp. + 12 microfiches. tunnicliff, s. p. 1980. A catalogue of the Lower Palaeozoic fossils in the collection of Major-General J. E. Portlock, R.E. , LL.D., F.R.S., E.G.S. etc. Ulster Museum, Belfast, 112 pp. williams, d. M. 1988. An illustrated catalogue of the type specimens in the Greville diatom herbarium. Bulletin of the Natural History Museum, Botany Series, 18, 1-148. wyse jackson, p. N. and monaghan, n. T. 1995. Transfer of the Huxley and Wright (1867) Carboniferous amphibian and fish material to Trinity College Dublin from the National Museum of Ireland. Journal of Paleontology, 69, 602-603. — and sleeman, A. G. 1990. Return of type, figured, referred and other fossils from the geological collections of Trinity College Dublin to the Geological Survey of Ireland. Bulletin of the Geological Survey of Ireland , 4, 243-244. INDEX The tripartite arrangement of the index (Taxonomic, Stratigraphical, Museums) follows that of Bassett (1975). In the taxonomic and museums indexes, the date of a publication is not given after the authors' names in cases where those authors have only one publication under their name in the bibliography; both names and dates are given in all other cases. Major taxonomic groupings are generally at Phylum level, or within commonly employed classificatory divisions that will be immediately familiar to palaeontologists. Where authors of catalogues have not themselves separated their specimens into the groupings adopted here, their genera and species are included as undifferentiated members of the highest appropriate division listed. The stratigraphical index is generally broken down to the level of geological Systems. In the Museums index, italicized information in square brackets draws attention to formal changes that have taken place in institutional names, and also to cases where material may have been transferred from one institution to another subsequent to its initial listing or description in publications. 602 PALAEONTOLOGY, VOLUME 40 TAXONOMIC INDEX INVERTEBRATA AMMONOIDEA: See MOLLUSCA [Ammonoidea] AMPHINEURA: See MOLLUSCA [Amphineura] ANNELIDA: \See also PROBLEMATICA, SYM- BIOSIS] Undifferentiated: Ordovician : Benton; Owens and Bassett; Tunni- cliff Silurian: Smith 1989, 1996 devonian: Benton carboniferous: Campbell; Rolfe et at. Jurassic: Pyrah 1976; Rolfe et al. cenozoic: Benton holocene: Benton Scolecodonts: Ordovician: Nudds 1983 carboniferous: Nudds 1988 ANTHOZOA: See CN1DARIA [AnthozoaJ ARACHNIDA: See ARTHROPODA [Chelicerata: Arachnida] ARCHAEOCYATHA: See PORIFERA (Archaeoeyathaj ARTHROPODA: Undifferentiated: age not indicated: Benton and Trewin Ordovician : Rolfe et al. Silurian : Rolfe et al. devonian : Rolfe et al. carboniferous : Rolfe et al. triassic : Rolfe et al. cretaceous : Rolfe et al. eocene : Rolfe et al. pliocene : Rolfe et al. quaternary: Rolfe et al. pleistocene : Rolfe et al. Chelicerata jLIndifferentiated]: Silurian: Nudds 1992a carboniferous: Nudds 1992a Chelicerata |Arachnida]: Silurian: Morris 1980 old red sandstone: Morris 1980 devonian: Morris 1980 carboniferous : Benton ; Hancock et al. ; Morris 1980; Owens and Bassett; Steward and Tor- rens; Strachan eocene: Morris 1980 oligocene: Morris 1980 miocene: Morris 1980 pleistocene: Morris 1980 Chelicerata |Merostomata]: carboniferous: Benton Crustacea [Undifferentiated]: age not indicated: Pyrah 1976 Silurian: Parkes and Sleeman carboniferous: Mitchell; Nudds 1992a; Parkes and Sleeman; Strachan triassic: Nudds 1992a Jurassic: Pyrah 1976 cretaceous: Nudds 1992a; Pyrah 1976 cenozoic: Pyrah 1976 holocene: Nudds 1983 Crustacea [Branchiopodal : old red sandstone: Morris 1980 devonian: Morris 1980 carboniferous: Morris 1980 permo-carboniferous: Morris 1980 Permian: Morris 1980 permo-triassic : Morris 1980 triassic: Morris 1980 Jurassic: Morris 1980 jurassic/lower cretaceous: Morris 1980 cretaceous: Morris 1980 oligocene: Morris 1980 Crustacea [Cirripedia): Silurian: Strachan Jurassic: Torrens 1978a cretaceous: Morris 1980 cenozoic: Morris 1980 eocene: Morris 1980 miocene: Morris 1980 pliocene: Morris 1980 pleistocene: Morris 1980 holocene: Morris 1980 Crustacea [Cycloidea]: carboniferous : Campbell ; Rolfe et al. Crustacea [Malacostraca] : Cambrian: Morris 1980 Ordovician: Morris 1980 Silurian: Benton; Morris 1980 devonian: Butler; Morris 1980 carboniferous: Morris 1980; Owens and Bassett permian: Morris 1980 triassic: Morris 1980 Jurassic: Morris 1980; Torrens 1978a cretaceous: Morris 1980 paleocene: Morris 1980 eocene: Morris 1980 oligocene: Morris 1980 miocene: Morris 1980 pliocene: Morris 1980 pleistocene: Morris 1980 Crustacea [Ostracoda|: age not indicated : Rolfe et al. Ordovician: Benton; Owens and Bassett DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 603 Silurian: Benton; Owens and Bassett devonian: Butler carboniferous: Campbell; Rolfe et al. Jurassic: Adams; Hodgkinson cretaceous: Hodgkinson cenozoic: Hodgkinson paleogene: Hodgkinson neogene: Hodgkinson miocene: Hodgkinson pleistocene: Hodgkinson holocene: Hodgkinson Crustacea |Phyllocarida|: Silurian: Campbell; Rolfe et al. ; Smith 1989 carboniferous: Campbell; Rolfe et at. Eurypterida: Cambrian: Morris 1980 Silurian: Campbell; Morris 1980; Nudds 1983; Rolfe et al. old red sandstone: Crane 1980a; Morris 1980 devonian: Morris 1980 carboniferous: Campbell; Crane 1980a; Morris 1980; Owens and Bassett; Rolfe et al. Insecta: carboniferous: Nudds 1983; Nudds 1992 a; Owens and Bassett triassic: Nudds 1992a; Pyrah 1976 cretaceous: Radley Myriapoda: old red sandstone: Morris 1980 carboniferous: Morris 1980 pleistocene: Morris 1980 Synxiphosura: Silurian: Morris 1980 Trilobita: age not indicated : Baird Cambrian: Morris 1988; Morris and Fortey; Owens and Bassett tremadoc: Rolfe et al.; Strachan Ordovician: Crane 1980a; Morris 1988; Morris and Fortey; Nudds 1983, 1988, 1992a; Owens and Bassett; Parkes and Sleeman; Rolfe et al.; Tripp and Howells; Tunnicliff Silurian: Chandler and Hannah; Morris 1988; Morris and Fortey; Nudds 1983, 1988; Owens and Bassett; Parkes and Sleeman; Rolfe et al.; Smith 1989, 1996; Strachan; Tripp and Howells devonian: Butler; Morris 1988; Morris and Fortey; Owens and Bassett; Rolfe et al. carboniferous: Campbell; Eagar and Preece; Morris 1988; Morris and Fortey; Nudds 1983, 1988, 1992a; Owens and Bassett; Parkes and Sleeman; Rolfe et al. permian : Owens and Bassett Trilobitomorpha: Ordovician: Benton Silurian: Benton devonian: Benton Xiphosura: Silurian: Morris 1980 carboniferous: Morris 1980; Owens and Bassett permian: Morris 1980 cretaceous: Morris 1980 ASTEROIDEA: See EC'HINODERMATA |Asteroidea] ASTEROZOA See ECHINODERMATA |Asterozoa[ BELEMNOIDEA: See MOLLUSCA [Belemnoidea| BIVALVIA: See MOLLUSCA [Bivalvial BLASTOIDEA: See ECHINODERMATA |BIastoidea] BRACHIOPODA: [Sec also SYMBIOSIS] age not indicated: Benton and Trewin; Pyrah 1976; Torrens and Taylor lower palaeozoic: Wyse Jackson and Sleeman Cambrian: Cocks; Owens and Bassett tremadoc : Rolfe et al. ; Strachan Ordovician: Benton; Cocks; Nudds 19826, 1992a; Owens and Bassett; Parkes and Slee- man; Rolfe et al.; Tunnicliff Silurian: Benton; Chandler and Hannah; Cocks; Nudds 19826; Owens and Bassett; Parkes and Sleeman; Rolfe et al.; Smith 1989, 1996; Torrens 1979; Tunnicliff old red sandstone: Smith 1989 devonian: Benton; Butler; Owens and Bassett; Rolfe et al. carboniferous: Campbell; Cocks; Mancenido and Damborenea; Mitchell; Nudds 19826, 1992a; Owens and Bassett; Parkes and Slee- man; Pyrah 1976, 19796; Rolfe et al. permian: Nudds 1992a; Pattison triassic: Rolfe et al. Jurassic: Mancenido and Damborenea; Owens and Bassett; Pyrah 1976; Rolfe et al. cretaceous: Boyd 1983; Pyrah 1976; Rolfe et al. cenozoic: Pyrah 1976 miocene: Rolfe et al. holocene: Benton BRANCH IOPODA: See ARTHROPODA [Crustacea: Branchiopoda] BRYOZOA: age not indicated: Benton and Trewin; Hancock et al. Ordovician: Benton; Eagar and Preece; Nudds 1992a; Owens and Bassett; Tunnicliff Silurian : Benton ; Chandler and Hannah ; Eagar and Preece; Loeffler and Crane; Nudds 1992a; Owens and Bassett; Rolfe et al. ; Smith 1989 604 PALAEONTOLOGY, VOLUME 40 devonian: Benton; Butler carboniferous: Benton; Campbell; Eagar and Preece; Loeffler and Crane; Nudds 1982a; Owens and Bassett; Parkes and Sleeman; Rolfe et al. permo-carboniferous : Benton Permian: Pattison Jurassic: Loeffler and Crane; Owens and Bas- sett; Pyrah 1976 Miocene: Rolfe et al. pliocene: Rolfe et al. quaternary: Pyrah 1976 pleistocene: Rolfe et al. holocene: Benton CALCICHORDATA: Cambrian : Owens and Bassett Ordovician : Owens and Bassett CEPHALOPODA: See MOLLUSCA [Cephalopoda] CHELICERATA: See ARTHROPODA [Chelicerata] CHITINOZOA: Silurian: Nudds 1983, 1988 CIRRSPEDIA: See ARTHROPODA [Crustacea: Cirripedia] CNIDARIA: Undifferentiated: age not indicated: Nudds 1982c precambrian: Nudds 1988 Ordovician : Rolfe et al. Silurian : Parkes and Sleeman ; Rolfe et al. devonian : Rolfe et al. carboniferous: Nudds 1982a, 1988; Parkes and Sleeman; Pyrah 1976; Rolfe et al. triassic : Rolfe et al. Jurassic: Pyrah 1976; Rolfe et al. cretaceous; Pyrah 1976; Rolfe et al. cenozoic: Pyrah 1976 eocene: Rolfe et at. oligocene : Rolfe et al. miocene : Rolfe et al. pliocene: Rolfe et al. quaternary: Pyrah 1976 pleistocene : Rolfe et al. Anthozoa [Undifferentiated|: ordovician: Tunnicliff Silurian: Smith 1989, 1996 carboniferous: Owens and Bassett; Sutherland permian: Pattison Jurassic : Loeffler and Crane Anthozoa [Heterocora!lia|: age not indicated: Benton and Trewin carboniferous: Benton; Campbell; Rolfe et al. Anthozoa [Octocorallia): age not indicated: Benton and Trewin holocene: Benton Anthozoa [Rugosa]: age not indicated: Benton and Trewin ordovician: Benton Silurian: Benton devonian: Benton carboniferous: Benton; Campbell; Rolfe et ai Anthozoa |Scleractinia[: age not indicated; Benton and Trewin holocene: Benton Anthozoa [Tabulata]: age not indicated: Benton and Trewin ordovician: Benton Silurian: Benton devonian: Benton carboniferous: Benton Anthozoa [Zoantharia): Silurian : Loeffler and Crane carboniferous: Loeffler and Crane Jurassic : Loeffler and Crane Cyclozoa: precambrian: Owens and Bassett Hydrozoa: ordovician: Tunnicliff triassic: Benton pliocene: Benton holocene: Benton Scyphozoa: lower palaeozoic: Loeffler and Crane ordovician: Tunnicliff Silurian: Tunnicliff; Smith 1989 carboniferous: Loeffler and Crane COELENTERATA: See CNIDARIA COLEOIDEA: See MOLLUSCA [Coleoidea] CONODONTA: See VERTEBRATA [CONODONTA| CONULARIIDA: See CNIDARIA [Scyphozoa| CONULATA: See CNIDARIA [Scyphozoa] CRICOCONARIDA: Tentaculitida: Silurian: Smith 1989 devonian: Butler CRINOIDEA: See ECHINODERMATA [Crinoidea| CRUSTACEA: See ARTHROPODA [Crustacea! CYCLOIDEA: See ARTHROPODA [Crustacea: Cycloidea| CYCLOZOA: See CNIDARIA [Cyclozoa] CYSTOIDEA: See ECHINODERMATA [Cystoidea] DENDROIDEA: See GRAPTOLITHINA |Dendroidea| DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 605 ECHINODERMATA |Undifferentiated|: age not indicated: Benton and Trewin Ordovician: Benton; Parkesand Sleeman; Rolfe et al. Silurian: Benton; Pyrah 1977; Parkes and Slee- man; Rolfe et al. carboniferous: Benton; Parkes and Sleeman; Pyrah 1977; Rolfe et al. Permian: Pattison Jurassic: Pyrah 1977; Rolfe et al. cretaceous : Rolfe et al. cenozoic: Pyrah 1977 eocene: Rolfe et al. Miocene: Rolfe et al. pliocene : Rolfe et al. quaternary: Pyrah 1977 pleistocene: Rolfe et al. holocene: Benton Asteroidea: Ordovician: Lewis 1993 Silurian: Lewis 1993 devonian: Lewis 1993 carboniferous: Lewis 1993 triassic: Lewis 1993 Jurassic: Lewis 1993 cretaceous: Lewis 1993 paleocene: Lewis 1993 eocene: Lewis 1993 Asterozoa [LIndifferentiated]: Ordovician: Nudds 1992a Silurian: Nudds 1992a carboniferous: Nudds 1983, 1988 Jurassic: Nudds 1983 Blastoidea : devonian: Butler carboniferous: Eagar and Preece; Nudds 1983, 1988, 1992a Crinoidea: Ordovician: Nudds 1988, 1992a; Owens and Bassett; Tunnicliff Silurian: Chandler and Hannah; Owens and Bassett; Smith 1989, 1996; Strachan devonian: Butler carboniferous: Nudds 1983, 1988; Phillips, P. W. permian: Nudds 1983 Jurassic: Campbell; Owens and Bassett; Phil- lips, P. W.; Rolfe et al. Cystoidea: tremadoc: Strachan Ordovician: Nudds 1988; Owens and Bassett; TunniclilT Echinoidea: Ordovician: Lewis 1986 Silurian: Lewis 1986 carboniferous: Lewis 1986; Nudds 1983; Phil- lips, P. W.; Rolfe et al. permian: Lewis 1986 triassic: Lewis 1986 Jurassic: Lewis 1986; Strachan; Torrens 1978a cretaceous: Ensom; Lewis 1986; Owens and Bassett cenozoic: Lewis 1986 paleocene: Lewis 1986 eocene: Lewis 1986 oligocene: Lewis 1986 oligocene-miocene : Lewis 1986 Miocene: Lewis 1986; Nudds 1983 pliocene: Lewis 1986 pleistocene: Lewis 1986 Holothuroidea: Jurassic: Hodgkinson Ophiuroidea: Ordovician: Lewis 1993 Silurian: Lewis 1993 devonian: Lewis 1993 carboniferous: Lewis 1993 triassic: Lewis 1993 Jurassic: Lewis 1993 cretaceous: Lewis 1993 eocene: Lewis 1993 Parablastoidea: Ordovician : Owens and Bassett Stelleroidea: Jurassic: Boyd 1983 ECHINOIDEA: See ECHINODERMATA [Echinoidea) EURYPTERIDA: See ARTHROPODA [Eurypterida| FORAMINIFERIDA: See PROTOZOA [Foraminiferidal GASTROPODA: See MOLLUSCA [Gastropoda) GONIATITINA: See MOLLUSCA [Ammonoidea) GRAPTOLITH1NA: [Undifferentiated]: age not indicated: Benton and Trewin tremadoc : Rolfe et al. Ordovician: Campbell; Owens and Bassett; Parkes and Sleeman; Rolfe et al . ; Tunnicliff Silurian : Campbell ; Owens and Bassett ; Parkes and Sleeman; Rolfe et al.'. Smith 1989, 1996; Tunnicliff Dendroidea: Ordovician: Benton Silurian: Benton Graptoloidea : Ordovician: Benton; Nudds 1983, 1988 Silurian: Benton; Nudds 1983, 1988 GRAPTOLOIDEA: See GRAPTOLI I HINA [Graptoloidea] 606 PALAEONTOLOGY, VOLUME 40 HETEROCORALLIA: See CNIDARIA |Anthozoa: Heterocorallia) HOLOTHUROIDEA: See ECHINODERMATA HYDROZOA: See CNIDARIA |Hydrozoal HYOLITHA: Ordovician : Parkes and Sleeman ICHNOFOSSILS: See also VERTEBRATE FOOT- PRINTS age not indicated: Benton and Trewin; Rolfe et al. precambrian: Owens and Bassett Cambrian: Nudds 1983; Parkes and Sleeman Ordovician: Benton; Nudds 1983; Owens and Bassett; Tunnicliff Silurian: Benton; Owens and Bassett; Smith 1989, 1996 OLD RED sandstone : Rolfe et al. carboniferous: Nudds 1992a; Rolfe et al. Permian: Nudds 1992a triassic: Nudds 1992a; Owens and Bassett INCERTAE SEDIS: See (PROBLEMATICAI INSECTA: See ARTHROPODA [lnsecta| LAMELLIBRANCHIA: See MOLLUSC A |Bivalvia| MACHAERIDIA: See PROBLEMATICA [Machaeridia] MALACOSTRACA: See ARTHROPODA [Crustacea: Malacostraca] MEDUSOIDS: See CNIDARIA [Cyclozoal MEROSTOMATA: See ARTHROPODA [Chelicerata: Merostomata] MOLLUSCA: Undifferentiated: pliocene: Garrad Ammonoidea : (including Goniatitina) devonian: Butler; Phillips, D. 1982a carboniferous: Campbell; Phillips, D. 1982a Permian: Phillips, D. 1982a triassic: Phillips, D. 1977 Jurassic: Boyd 1983; Crane 1980a; Knell; Phil- lips, D. 1977; Powell and Edmonds; Torrens 19786 cretaceous: Boyd 1983; Phillips, D. 1977; Radley Amphineura: carboniferous: Campbell; Rolfe et al. Permian: Pattison quaternary: Pyrah 1978 Belemnoidea: Jurassic: Powell and Edmonds Bivalvia: [See also PROBLEMATICA, SYMBI- OSIS) age not indicated: Benton and Trewin Ordovician: Benton; Nudds 19826, 1988, 1992a; Owens and Bassett; Rolfe et al.; Tunnicliff Silurian: Benton; Nudds 19826; Owens and Bassett; Rolfe et al.; Smith 1989, 1996; Tunnicliff old red sandstone: Smith 1989, 1996 devonian: Butler; Parkes and Sleeman carboniferous: Benton; Campbell; Chandler and Hannah; Eagar and Preece; Nudds 19826, 1988, 1992a, 19926; Owens and Bassett; Parkes and Sleeman; Pattison; Pyrah 1977; Rolfe et al. permian: Eagar and Preece; Nudds 1992a; Pat- tison; Rolfe et al. triassic: Owens and Bassett; Rolfe et al. Jurassic: Boyd 1983; Crane 1980a; Eagar and Preece; Mancenido and Damborenea; Nudds 1992a; Owens and Bassett; Parkes and Slee- man; Pyrah 1977; Rolfe et al.; Strachan; Torrens 19786 cretaceous: Benton; Pyrah 1977; Radley; Rolfe et al. cenozoic: Pyrah 1977 eocene : Rolfe et al. oligocene: Phillips, P. W.; Rolfe et al. Miocene: Rolfe et al. pliocene: Rolfe et al.; Torrens 1983 quaternary: Pyrah 1977 pleistocene: Rolfe et al. holocene: Owens and Bassett; Phillips, P. W.; Rolfe et al. Cephalopoda [Undifferentiated|: age not indicated: Benton and Trewin Ordovician: Benton; Owens and Bassett; Parkes and Sleeman; Phillips, D. 19826; Rolfe et al.; Tunnicliff Silurian: Owens and Bassett; Parkes and Slee- man; Phillips, D. 19826; Smith 1989, 1996 old red sandstone: Smith 1989 devonian: Phillips, D. 19826 carboniferous: Nudds 19826, 1988, 1992a; Owens and Bassett; Parkes and Sleeman; Phillips, D. 19826; Pyrah 1978; Rolfe et al. permian: Pattison triassic: Phillips, D. 19826; Rolfe et al. Jurassic: Clark; Nudds 1992a; Owens and Bas- sett; Phillips, D. 19826; Pyrah 1978; Rolfe et al.; Strachan cretaceous: Nudds 1992a; Owens and Bassett; Phillips, D. 19826; Pyrah 1978; Rolfe et al. eocene : Rolfe et al. miocene: Phillips, D. 19826 Coleoidea: permian: Phillips, D. 1982a DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 607 triassic: Phillips, D. 1982a Jurassic: Phillips, D. 1982a cretaceous: Phillips, D. 1982a eocene: Phillips, D, 1982a Gastropoda : age not indicated: Benton and Trewin tremadoc: Rolfe et al. Ordovician: Benton; Nudds 19826; Owens and Bassett; Parkes and Sleeman; Rolfe el al.\ Tunnicliff Silurian: Benton; Owens and Bassett; Rolfe et al.\ Smith 1989, 1996; Tunnicliff old red sandstone: Smith 1989 devonian: Benton; Butler carboniferous: Campbell; Mitchell; Nudds 19826, 1992a; Owens and Bassett; Parkes and Sleeman; Phillips, P. W.; Pyrah 1978; Rolfe et al. permian: Nudds 1992a; Pattison triassic: Rolfe et al. Jurassic: Pyrah 1978; Rolfe et al. cretaceous: Pyrah 1978; Radley; Rolfe et al. cenozoic : Rolfe et al. paleogene: Radley eocene: Pyrah 1978; Rolfe et al. oligocene : Rolfe et al. miocene: Rolfe et al. pliocene: Pyrah 1978; Rolfe et al. quaternary: Pyrah 1978 pleistocene: Nudds 1992a; Owens and Bassett; Rolfe et al. Nautiloidea: [See also SYMBIOSIS] Ordovician: Phillips, D 1982a; Tunnicliff Silurian: Phillips, D. 1982a; Tunnicliff devonian: Butler; Phillips, D. 1982a carboniferous: Eagar and Preece; Phillips, D. 1982a Permian: Phillips, D. 1982a triassic: Phillips, D. 1982a Jurassic: Phillips, D. 1982a; Torrens and Taylor cretaceous: Phillips, D. 1982a eocene: Phillips, D. 1982a miocene: Phillips, D. 1982a Rostroconchia: Ordovician: Benton Silurian : Parkes and Sleeman carboniferous: Parkes and Sleeman Scaphopoda: carboniferous: Campbell; Rolfe et al. permian: Nudds 1992a; Riley quaternary: Pyrah 1978 MYRIAPODA: See ARTHROPODA |Myriapoda| NAUTILOIDEA: See MOLLUSCA |Nautiloidea| OC'TOCORALLI A : See CNIDAR1A |Anthozoa: Octocorallia] OPHIUROIDEA: See ECHINODERMATA |Ophiuroidea| OSTRACODA: See ARTHROPODA |Crustacea: Ostracoda| P AR ABLASTOI DEA See ECHINODERMATA [ParabIastoidea| PELECYPODA: See MOLLUSCA |Bivalvia] PHYLLOCARIDA: See ARTHROPODA [Crustacea: Phyllocarida| POLYPLACOPHORA: See MOLLUSCA [Amphineura] PORIFERA: [Undifferentiated|: age not indicated: Benton and Trewin Cambrian: Nudds 1982a; Owens and Bassett Ordovician: Owens and Bassett; Rolfe et al. ; Tunnicliff Silurian: Benton; Rolfe et al.', Tunnicliff carboniferous: Campbell; Nudds 1982a, 1988; Rolfe et al. permian: Pattison Jurassic: Pyrah 1976; Torrens and Taylor cretaceous: Loeffler and Crane; Nudds 1982a; Pyrah 1976 Archaeocyatha: Cambrian: Nudds 1982a Stromatoporoidea : Ordovician: Benton; Tunnicliff Silurian: Benton; Loeffler and Crane devonian: Benton Jurassic: Loeffler and Crane PROBLEM ATICA: [See also VERTEBRATA PROBLEMATIC^ and PLANTAE PROBLEM- ATICA] Llmdifferentiated: precambrian: Benton Cambrian: Nudds 1988 Ordovician: Morris 1980; Nudds 1983 Silurian: Morris 1980; Nudds 1983, 1988; Owens and Bassett; Parkes and Sleeman carboniferous: Nudds 1983 7ANNELIDA: Ordovician: Benton Silurian: Benton devonian: Benton MACHAERIDIA: Ordovician: Benton PERFORATIONS MADE BY BIVALVES: permian: Pattison PHYLUM UNCERTAIN: age not indicated: Benton and Trewin SMALL CONOIDAL SHELLS OF UNCER- TAIN AFFINITY: Ordovician: Benton devonian: Benton 608 PALAEONTOLOGY, VOLUME 40 PROTOZOA: Foraminiferida : age not indicated: Murray and Taplin Ordovician: Benton carboniferous : Adams et al. ; Campbell ; Nudds 1982a, 1988; Rolfe et al. Permian : Rolfe et al. triassic: Adams et al. Jurassic: Adams; Owens and Bassett; Rolfe et al. cretaceous: Adams et al.; Benton cenozoic: Adams et al. eocene: Adams et al.; Benton; Rolfe et al. miocene : Adams et al. ; Rolfe et al. pliocene: Rolfe et al. holocene: Adams; Adams et al.; Benton; Joysey Radiolaria: Jurassic : Rolfe et al. RADIOLARIA: See PROTOZOA |Radiolaria| ROSTROCONCHIA: See MOLLUSCA |Rostroconchia| RUGOSA: See CNIDARIA (Anthozoa: Rugosa) SCAPHOPODA: See MOLLUSCA [Scaphopodaj SCLERACTINIA: See CNIDARIA [Anthozoa: Scleractiniaj SCOLECODONTS: See ANNELIDA [Scolecodonts] SCYPHOZOA: See CNIDARIA [Scyphozoaf STELLEROIDEA: See ECHINODERMATA [Stelleroidea] STROMATOPOROIDEA : See PORIFERA [Stromatoporoideaj SYMBIOSIS: Annelida/Brachiopoda : Silurian: Nudds 1983 MoIIusca |Bivalvia]/Brachiopoda: Silurian: Nudds 1983 MoIIusca [NautiloideaJ/Annelida: Silurian: Nudds 1983 MoIIusca [Nautiloidea]/Brachiopoda : Silurian: Nudds 1983, 1988 SYNXIPHOSURA: See ARTHROPODA [Synxiphosura| TABULATA: See CNIDARIA | Anthozoa: Tabulata] TENTACULITIDA: See CRICOCONARIDA [Tentaculitida] TRACE FOSSILS: See ICHNOFOSSILS TRILOBITA: See ARTHROPODA [Trilobita] TRILOBITOMORPHA : See ARTHROPODA |Trilobitomorpha] UNKNOWN: See PROBLEMATICA XIPHOSURA: See ARTHROPODA [Xiphosural ZOANTHAR1A: See CNIDARIA [Anthozoa: Zoanthariaj vertebrata UNDIFFERENTIATED: Jurassic: Torrens 19786 AMPHIBIA: carboniferous: Boyd 1986; Boyd and Turner; Nudds 1983, 1992a; Rolfe et al.; Wyse Jack- son and Monaghan AVES: cretaceous: Pyrah 1979a quaternary: Pyrah 1979a CONODONTA: Ordovician: Nudds 1988; Owens and Bassett Silurian : Owens and Bassett carboniferous: Nudds 1983, 1988; Owens and Bassett; Rolfe et al. MAMMALIA: AGE NOT indicated : Rolfe et at. triassic: Duffin 1978 Jurassic: Pyrah 1979a cenozoic: Pyrah 1979a Paleogene: Radley mio-pliocene : Rolfe et al. quaternary: Campbell; Pyrah 1979a; Rolfe et al. pleistocene: Boyd 1983; Nudds 1983, 1988, 1992a; Radley; Rolfe et al. PISCES: age not indicated : Andrews Silurian: Campbell; Rolfe et al.; Smith 1989 old red sandstone: Campbell; Nudds 1992a; Paton 1976, 1981; Pyrah 1979a; Rolfe et al.; Smith 1989, 1996 devonian; Butler; Nudds 1983, 1992a; Owens and Bassett; Paton 1981 ; Rolfe et al. carboniferous: Campbell; Nudds 1992a; Owens and Bassett; Parkes and Sleeman; Paton 1976; Pyrah 1979a; Rolfe et al.; Wyse Jackson and Monaghan permian: Paton 1976; Pyrah 1979a triassic: Duffin 1978; Nudds 1992a; Owens and Bassett DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 609 Jurassic: Boyd 1983; Cross 1975a, 1 9756 ; Duffin 1978; Powell and Edmonds; Pyrah 1979a; Torrens 19786 cretaceous: Pyrah 1979a cenozoic: Pyrah 1979a eocene: Nudds 1983 pliocene; Rolfe et al. pleistocene : Rolfe et al. holocene: Nudds 1992a PROBLEMATICA: [See also 1NVERTEBRATA PROBLEMATICA and PLANTAE PROBLEM- ATICA] age not indicated: Benton and Trewin REPTILIA: age not indicated: Benton and Trewin devonian : Parkes and Sleeman Permian: Rolfe et al. triassic: Benton and Trewin; Duffin 1978; Nudds 1992a; Rolfe et al. Jurassic: Boyd 1983; Cross 1975a, 19756; Duffin 1978, 1979; Nudds 1983, 1992a; Owens and Bassett; Powell and Edmonds; Pyrah 1979a; Rolfe et al. ; Torrens 19786 cretaceous: Crane and Getty; Pyrah 1979a; Radley; Rolfe et al. paleogene: Radley pliocene: Rolfe et al. holocene : Rolfe et al. VERTEBRATE FOOTPRINTS: carboniferous: Sarjeant Permian: Rolfe et al.; Sarjeant triassic: Sarjeant Jurassic: Pyrah 1979a; Rolfe et al.; Sarjeant cretaceous: Radley; Sarjeant PLANTAE UNDIFFERENTIATED: Silurian: Nudds 1989 old red sandstone: Nudds 1984, 1992a devonian: Crane 19806; Nudds 1984; Parkes and Sleeman carboniferous: Crane 19806; Eagar and Preece; Hancock et a!. ; Newman and Chatt-Ramsey ; Nudds 1984, 1989, 1992a; Parkes and Sleem- an; Phillips, P. W.; Pyrah 1979a; Strachan permian: Nudds 1992a; Pattison triassic: Crane 19806; Phillips, P. W. Jurassic: Crane 19806; Newman and Chatt- Ramsey; Nudds 1992a; Powell and Ed- monds: Pyrah 1979a; Torrens 19786 cretaceous: Crane 19806; Hancock et al. cenozoic: Hancock et al. paleogene: Parkes and Sleeman oligocene: Crane and Getty ACRITARCHA: Cambrian: Nudds 1984 Ordovician: Nudds 1984 Silurian: Nudds 1984, 1989 carboniferous: Nudds 1984 ALGAE: [See also PLANTAE PROBLEMATICA] age not indicated: Benton and Trewin Cambrian: Crawley Ordovician: Benton; Crawley Silurian: Crawley; Owens and Bassett; Smith 1989 old red sandstone: Crawley devonian: Crawley carboniferous: Campbell; Crawley; Owens and Bassett; Rolfe et al. permian: Crawley triassic: Crawley Jurassic: Crawley cretaceous: Crawley paleogene: Crawley eocene: Crawley oligocene: Crawley miocene: Crawley pliocene: Crawley pleistocene: Crawley holocene: Crawley ANGIOSPERMAE: See TRACHEOPHYTA | Angiospermae] DIATOMS: (BACILLARIOPHYTA) eocene: Williams oligocene: Williams holocene: Williams MEGASPORES: devonian : Parkes and Sleeman MIOSPORES: Cambrian: Nudds 1984 Ordovician: Nudds 1984 Silurian: Nudds 1984, 1989 devonian: Nudds 1984, 1989; Parkes and Slee- man carboniferous: Nudds 1984, 1989; Parkes and Sleeman Jurassic: Parkes and Sleeman PROBLEMATICA: [See also INVERTEBRATA PROBLEMATICA and VERTEBRATA PROB- LEMATICA] Silurian: Owens and Bassett: Parkes and Slee- man devonian: Owens and Bassett ?Algae age not indicated : Rolfe et al. PTERIDOPHYTA: devonian : Rolfe et al. carboniferous: Campbell; Rolfe et al; Wyse Jackson and Sleeman 610 PALAEONTOLOGY, VOLUME 40 TRACHEOPHYTA: Undifferentiated: Silurian : Owens and Bassett devonian : Owens and Bassett carboniferous: Owens and Bassett Permian: Owens and Bassett triassic: Owens and Bassett cretaceous/tertiary: Owens and Bassett Angiospermae: cenozoic: Rolfe et al. ?paleocene: Campbell STRATIGRAPHICAL INDEX Age not indicated: See ALGAE, ARTHROPODA [Undifferentiated] [Crustacea: Undifferentiated] [Crustacea: Ostracoda] [Trilobita], BRACHIO- PODA, BRYOZOA, CNIDARIA [Undifferentia- ted] [Anthozoa: Heterocorallia] [Anthozoa: Octo- corallia] [Anthozoa: Rugosa] [Anthozoa: Sclerac- tinia] [Anthozoa: Tabulata], ECHINODER- MATA [Undifferentiated], ICHNOFOSSILS, GRAPTOLITHINA [Undifferentiated], MAM- MALIA, MOLLUSCA [Bivalvia] [Cephalopoda: Undifferentiated] [Gastropoda], PISCES, PLAN- TAE [Problematica], PORIFERA [Undifferentia- ted], PROBLEMATICA, PROTOZOA [Forami- niferida], REPTILIA, VERTEBRATA [Problem- atica] Precambrian: See CNIDARIA [Undifferentiated] [Cyclozoa], ICHNOFOSSILS, PROBLEMATICA [Undifferentiated] Lower Palaeozoic: See BRACHIOPODA, CNIDA- RIA [Scyphozoa] Cambrian: See ACRITARCHA, ALGAE, ARTH- ROPODA [Crustacea : Malacostraca] [Euryptenda] [Trilobita], BRACHIOPODA, CALCICHORD- ATA, ICHNOFOSSILS, MIOSPORES, PORIF- ERA [Undifferentiated] [Archaeocyatha], PROB- LEMATICA Tremadoc: See ARTHROPODA [Trilobita], BRACH- IOPODA, ECHINODERMATA [Cystoidea], GRAPTOLITHINA [Undifferentiated], MOL- LUSCA [Gastropoda] Ordovician: See ACRITARCHA, ALGAE, ANNE- LIDA [Undifferentiated] [Scolecodonts], ARTH- ROPODA [Undifferentiated] [Crustacea: Mala- costraca] [Crustacea: Ostracoda] [Trilobita] [Tri- lobitomorpha], BRACHIOPODA, BRYOZOA, CALCICHORDATA, CNIDARIA [Undifferenti- ated] [Anthozoa: Undifferentiated] [Anthozoa: Rugosa] [Anthozoa: Tabulata] [Hydrozoa] [Scy- phozoa], CONODONTA, ECHINODERMATA [Undifferentiated] [Asteroidea] [Asterozoa: Undif- ferentiated] [Crinoidea] [Cystoidea] [Echinoidea] [Ophiuroidea] [Parablastoidea] GRAPTOLITH- INA [Undifferentiated] [Dendroidea] [Grapto- loidea], HYOLITHA, ICHNOFOSSILS, MIO- SPORES, MOLLUSCA [Bivalvia] [Cephalopoda: Undifferentiated] [Gastropoda] [Nautiloidea] [Ros- troconchia], PROBLEMATICA, PROTOZOA, PORIFERA [Undifferentiated] [Stromatoporoidea] Silurian: See ACRITARCHA, ALGAE, ANNE- LIDA [Undifferentiated], ARTHROPODA [Undif- ferentiated] [Chelicerata: Undifferentiated] [Chel- icerata: Arachnida] [Crustacea: Undifferentia- ted] [Crustacea: Cirripedia] [Crustacea: Malaco- straca] [Crustacea: Ostracoda] [Crustacea: Phyllo- carida] [Eurypterida] [Synxiphosura] [Trilobita] [Trilobitomorpha] [Xiphosura], BRACHIOPODA, BRYOZOA, CHITINOZOA, CNIDARIA [Un- differentiated] [Anthozoa: Undifferentiated] [Anth- ozoa: Rugosa] [Anthozoa: Tabulata] [Anthozoa: Zoantharia] [Scyphozoa], CONODONTA, CRIC- OCONARIDA [Tentaculitida], ECHINODER- MATA [Undifferentiated] [Asteroidea] [Asterozoa: Undifferentiated] [Crinoidea] [Echinoidea] [Ophi- uroidea], GRAPTOLITHINA [Undifferentiated] [Dendroidea] [Graptoloidea], ICHNOFOSSILS, MIOSPORES, MOLLUSCA [Bivalvia] [Cepha- lopoda: Undifferentiated] [Gastropoda] [Nauti- loidea] [Rostroconchia], PISCES, PLANTAE [Undifferentiated] [Problematica] [Tracheophyta], PORIFERA [Undifferentiated] [Stromatoporoi- dea], PROBLEMATICA, SYMBIOSIS, TRACH- EOPHYTA [Undifferentiated] Old Red Sandstone: See ALGAE, ARTHROPODA [Chelicerata: Arachnida] [Crustacea: Branchi- opoda] [Eurypterida] [Myriapoda], BRACHIO- PODA, ICHNOFOSSILS, MOLLUSCA [Bival- via] [Cephalopoda: Undifferentiated] [Gastro- poda], PISCES, PLANTAE [Undifferentiated] Devonian: See ALGAE, ANNELIDA [Undifferent- iated], ARTHROPODA [Undifferentiated] [Chel- icerata: Arachnida] [Crustacea: Branchiopoda] [Crustacea: Malacostraca] [Crustacea: Ostracoda] [Eurypterida] [Trilobita] [Trilobitomorpha], BRACHIOPODA, BRYOZOA, CNIDARIA [Undifferentiated] [Anthozoa: Rugosa] [Anthozoa: Tabulata], CRICOCONARIDA [Tentaculitida], ECHINODERMATA [Asteroidea] [Blastoidea] [Crinoidea] [Ophiuroidea], MIOSPORES, MOL- LUSCA [Ammonoidea] [Bivalvia] [Cephalopoda: Undifferentiated] [Gastropoda] [Nautiloidea], PISCES, PLANTAE [Undifferentiated] [Mega- spores] [Miospores] [Problematica] [Tracheo- phyta], PORIFERA [Stromatoporoidea], PROB- DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 611 LEMATICA, PTER1DOPHYTA, REPTILIA, TRACHEOPHYTA [Undifferentiated] Carboniferous: See ACRITARCHA, ALGAE, AMPHIBIA, ANNELIDA [Undifferentiated] [Sco- lecodonts], ARTHROPODA [Undifferentiated] [Chelicerata : Undifferentiated] [Chelicerata: Arachnida] [Chelicerata : Merostomata] [Crustacea : Undifferentiated] [Crustacea: Branchiopoda] [Crustacea: Cycloidea] [Crustacea: Malacostraca] [Crustacea: Ostracoda] [Crustacea: Phyllocarida] [Eurypterida] [Insecta] [Myriapoda] [Trilobita] [Xiphosura], BRACHIOPODA, BRYOZOA, CNI- DARIA [Undifferentiated] [Anthozoa: Undifferen- tiated] [Anthozoa: Heterocorallia] [Anthozoa: Rugosa] [Anthozoa: Tabulata] [Anthozoa: Zoan- tharia] [Scyphozoa], CONODONTA, ECHINO- DERMATA [Undifferentiated] [Asteroidea] [Aster- ozoa: Undifferentiated] [Blastoidea] [Crinoidea] [Echinoidea] [Ophiuroidea], ICHNOFOSSILS, MIOSPORES, MOLLUSCA [Ammonoidea] [Am- phineura] [Bivalvia] [Cephalopoda: Undifferentia- ted] [Gastropoda] [Nautiloidea] [Rostroconchia] [Scaphopoda], PISCES, PLANTAE [Undifferen- tiated] [Miospores], PORIFERA [Undifferentia- ted], PROBLEMATICA, PROTOZOA [Foramini- ferida], PTERIDOPHYTA, TRACHEOPHYTA [Undifferentiated], VERTEBRATE FOOT- PRINTS Permo-Carboniferous: See ARTHROPODA [Crus- tacea: Branchiopoda], BRYOZOA Permian: See ALGAE, ARTHROPODA [Crus- tacea: Branchiopoda] [Crustacea: Malacostraca] [Trilobita] [Xiphosura], BRACHIOPODA, BRYO- ZOA, CNIDARIA [Anthozoa: Undifferentiated], ECHINODERMATA [Undifferentiated] [Crinoi- dea] [Echinoidea], ICHNOFOSSILS, MOL- LUSCA [Ammonoidea] [Amphineura] [Bivalvia] [Cephalopoda: Undifferentiated] [Coleoidea] [Gas- tropoda] [Nautiloidea] [Scaphopoda], PISCES, PLANTAE [Undifferentiated], PORIFERA [Un- differentiated], PROBLEMATICA, PROTOZOA [Foraminiferida], REPTILIA, TRACHEOPHYTA [Undifferentiated], VERTEBRATE FOOT- PRINTS Permo-Triassie: See ARTHROPODA [Crustacea: Branchiopoda] Triassic: See ALGAE, ARTHROPODA [Undiffer- entiated] [Crustacea: Undifferentiated] [Crustacea: Branchiopoda] [Crustacea : Malacostraca] [Insecta], BRACHIOPODA, CNIDARIA [Undifferenti- ated] [Hydrozoa], ECHINODERMATA [Asteroi- dea] [Echinoidea] [Ophiuroidea], ICHNOFOS- SILS, MAMMALIA, MOLLUSCA [Ammonoi- dea] [Bivalvia] [Cephalopoda: Undifferentiated] [Coleoidea] [Gastropoda] [Nautiloidea], PISCES, PLANTAE [Undifferentiated], PROTOZOA [For- aminiferida], TRACHEOPHYTA [Undifferentia- ted], REPTILIA, VERTEBRATE FOOTPRINTS Jurassic: See ALGAE, ANNELIDA [Undifferentia- ted], ARTHROPODA [Crustacea: Undifferentia- ted] [Crustacea: Branchiopoda] [Crustacea: Cirri- pedia] [Crustacea: Malacostraca] [Crustacea: Ost- racoda], BRACHIOPODA, BRYOZOA, CNIDA- RIA [Undifferentiated] [Anthozoa: Undifferentia- ted] [Anthozoa: Zoantharia], ECHINODERMA- TA [Undifferentiated] [Asteroidea] [Asterozoa: Undifferentiated] [Crinoidea] [Echinoidea] [Holo- thuroidea] [Ophiuroidea] [Stelleroidea], MAMMA- LIA, MOLLUSCA [Ammonoidea] [Belemnoidea] [Bivalvia] [Cephalopoda: Undifferentiated] [Coleoi- dea] [Gastropoda] [Nautiloidea], PISCES, PLAN- TAE [Undifferentiated] [Miospores], PORIFERA [Undifferentiated] [Stromatoporoidea], PROTO- ZOA [Foraminiferida], PROTOZOA [Radiolaria], REPTILIA, VERTEBRATA [Undifferentiated], VERTEBRATE FOOTPRINTS Jurassic/Lower Cretaceous: See ARTHROPODA [Crustacea: Branchiopoda] Cretaceous: See ALGAE, ARTHROPODA [Undif- ferentiated] [Crustacea: Undifferentiated] [Crusta- cea: Branchiopoda] [Crustacea: Cirripedia] [Crus- tacea: Malacostraca] [Crustacea: Ostracoda], [Insecta] [Xiphosura], AVES, BRACHIOPODA, CNIDARIA [Undifferentiated], [ECHINODER- MATA [Undifferentiated] [Asteroidea] [Echinoi- dea] [Ophiuroidea], MOLLUSCA [Ammonoidea] [Bivalvia] [Cephalopoda; Undifferentiated] [Coleoi- dea] [Gastropoda] [Nautiloidea], PISCES, PLAN- TAE [Undifferentiated], PORIFERA [Undifferen- tiated], PROTOZOA [Foraminiferida], REPTI- LIA, VERTEBRATE FOOTPRINTS Cretaceous/Tertiary: See TRACHEOPHYTA [Un- differentiated] Cenozoic: See ANNELIDA [Undifferentiated], ARTHROPODA [Crustacea: Undifferentiated] [Crustacea: Cirripedia] [Crustacea: Ostracoda], BRACHIOPODA, CNIDARIA [Undifferentia- ted], ECHINODERMATA [Undifferentiated] [Echinoidea], MAMMALIA, MOLLUSCA [Bival- via] [Gastropoda], PISCES, PLANTAE [Undiffer- entiated], PROTOZOA [Foraminiferida], TRACH- EOPHYTA [Angiospermae] Paleogene: See ARTHROPODA [Crustacea: Ostra- coda], MAMMALIA, MOLLUSCA [Gastropoda], PLANTAE [Undifferentiated], REPTILIA Paleocene: See ALGAE, ARTHROPODA [Crusta- cea: Malacostraca], ECHINODERMATA [Aster- oidea] Echinoidea], TRACHEOPHYTA [Angio- spermae] Eocene: See ALGAE, ARTHROPODA [Undifferen- tiated] [Chelicerata: Arachnida] [Crustacea: Cirri- pedia] [Crustacea: Malacostraca], CNIDARIA [Undifferentiated], DIATOMS, ECHINODERM- 612 PALAEONTOLOGY, VOLUME 40 ATA [Undifferentiated] [Asteroidea] [Echinoidea] [Ophiuroidea], MOLLUSCA [Bivalvia] [Cephalo- poda: Undifferentiated] [Coleoidea] [Gastropoda] [Nautiloidea], PISCES, PROTOZOA [Foraminif- erida] Oligocene: See ALGAE, ARTHROPODA [Chelice- rata: Arachnida] [Crustacea: Branchiopoda] [Crus- tacea: Malacostraca], CNIDARIA [Undifferenti- ated], DIATOMS, ECHINODERMATA [Echinoi- dea], MOLLUSCA [Bivalvia] [Gastropoda], PLANTAE [Undifferentiated] Oligocene-Miocene: See ECHINODERMATA [Echinoidea] Neogene: ARTHROPODA [Crustacea: Ostracoda] Miocene: See ALGAE, ARTHROPODA [Chelicer- ata: Arachnida] [Crustacea: Cirripedia] [Crusta- cea: Malacostraca] [Crustacea: Ostracoda], BRA- CHIOPODA, BRYOZOA, CNIDARIA [Undiffer- entiated], ECHINODERMATA [Undifferentia- ted] [Echinoidea], MOLLUSCA [Bivalvia] [Cepha- lopoda: Undifferentiated] [Gastropoda] [Nauti- loidea], PROTOZOA [Foraminiferida] Mio-PIiocene: See MAMMALIA Pliocene: See ALGAE, ARTHROPODA [Undiffer- entiated] [Crustacea: Cirripedia] [Crustacea: Mala- costraca], BRYOZOA, CNIDARIA [Undifferen- tiated] [Hydrozoa], ECHINODERMATA [Undif- ferentiated] [Echinoidea], MOLLUSCA [Undiffer- entiated] [Bivalvia] [Gastropoda], PISCES, PRO- TOZOA [Foraminiferida], REPTILIA Quaternary: See ARTHROPODA [Undifferentia- ted], AVES, BRYOZOA, CNIDARIA [Undifferen- tiated], ECHINODERMATA [Undifferentiated], MAMMALIA, MOLLUSCA [Amphineura] [Bivalvia] [Gastropoda] [Scaphopoda] Pleistocene: See ALGAE, ARTHROPODA [Undif- ferentiated] [Chelicerata: Arachnida] [Crustacea: Cirripedia] [Crustacea: Malacostraca] [Crustacea: Ostracoda] [Myriapoda], BRYOZOA, CNIDA- RIA [Undifferentiated], ECHINODERMATA [Undifferentiated] [Echinoidea], MAMMALIA, MOLLUSCA [Bivalvia] [Gastropoda], PISCES Holocene: See ALGAE, ANNELIDA [Undifferen- tiated], ARTHROPODA [Crustacea: Undifferen- tiated] [Crustacea: Cirripedia] [Crustacea: Ostra- coda], BRACHIOPODA, BRYOZOA, CNIDA- RIA [Anthozoa: Octocorallia] [Anthozoa: Sclerac- tinia] [Hydrozoa], DIATOMS, ECHINODER- MATA [Undifferentiated], MOLLUSCA [Bival- via], PISCES, PROTOZOA [Foraminiferida], REPTILIA MUSEUMS INDEX ABERDEEN UNIVERSITY, GEOLOGY DE- PARTMENT: Benton; Benton and Trewin ABERYSTWYTH, UNIVERSITY COLLEGE OF WALES: Cocks ABINGDON MUSEUM : [ All type specimens trans- ferred to University Museum, Oxford ] Knell BANGOR, UNIVERSITY COLLEGE MUSEUM: Morris 1988 BARNSTAPLE: See MUSEUM OF NORTH DEVON, BARNSTAPLE BATH ROYAL LITERARY AND SCIENTIFIC INSTITUTION [BATH MUSEUM]: Crane 19806; Duffin 1978, 1979 BELFAST: See ULSTER MUSEUM, BELFAST BIRMINGHAM UNIVERSITY LAPWORTH MUSEUM: Cocks; Morris 1988; Smith 1989; Strachan BOLTON MUSEUM AND ART GALLERY: Hancock et al. BRISTOL CITY MUSEUM AND ART GAL- LERY: Crane 1980a, 19806; Loeffler and Crane; Morris 1988; Smith 1989; Torrens and Taylor BRISTOL UNIVERSITY GEOLOGY DEPART- MENT MUSEUM: Loeffler and Crane; Morris 1988 BRITISH GEOLOGICAL SURVEY, EDIN- BURGH: [See also note, p. 615 ] Benton; Clark; Cocks; Morris 1988 BRITISH GEOLOGICAL SURVEY, LONDON: [Formerly also Geological Museum of the British Geological Survey, London ; also Geological Museum, Institute of Geological Sciences, London ; also Geological Survey Museum , London. All mater- ial now transferred to British Geological Survey, Keyworth, Nottingham ( see note, p. 615)] Andrews; Benton; Clark; Cross 1975a, 19756; Hancock et al. \ Mitchell; Sarjeant; Tunnicliff BRITISH GEOLOGICAL SURVEY, KEY- WORTH, NOTTINGHAM: [See also above and note p. 615] Cocks; Morris 1988; Smith 1989, 1996 BRITISH MUSEUM (NATURAL HISTORY): See NATURAL HISTORY MUSEUM, LONDON CAMBRIDGE: See SEDGWICK MUSEUM, and UNIVERSITY OF CAMBRIDGE, ZOOLOGI- CAL MUSEUM CARDIFF: See NATIONAL MUSEUM OF WALES CARLISLE MUSEUM: See TULLIE HOUSE MUSEUM AND ART GALLERY, CARLISLE CASTLE MUSEUM, NORWICH: Sarjeant CENTRAL MUSEUM AND ART GALLERY, NORTHAMPTON: Smith 1989, 1996 CHELTENHAM ART GALLERY AND MUSEUM: Torrens 1978a; Torrens and Taylor CHELTENHAM COLLEGE MUSEUM [Specimens transferred to Bristol City Museum, Natural History DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 613 Museum, and Portsmouth Polytechnic (now Uni- versity of Portsmouth)} -. Torrens and Taylor CHESTER: See GROSVENOR MUSEUM, CHES- TER CLIFFE CASTLE MUSEUM, KEIGHLEY: Cocks DICK INSTITUTE, KILMARNOCK: Benton DOUGLAS, ISLE OF MAN: See MANX MUSEUM DUBLIN: See TRINITY COLLEGE, DUBLIN and NATIONAL MUSEUM OF IRELAND, DUB- LIN and GEOLOGICAL SURVEY OF IRE- LAND, DUBLIN DUDLEY ART GALLERY AND MUSEUM [in- cluding specimens formerly housed in Dudley Central Library ]: Chandler and Hannah; Morris 1988; Smith 1989; Torrens 1979 EDINBURGH: See BRITISH GEOLOGICAL SURVEY, EDINBURGH and GRANT INSTI- TUTE, EDINBURGH and NATIONAL MUSEUMS OF SCOTLAND, EDINBURGH and ROYAL SCOTTISH MUSEUM, EDINBURGH ELGIN MUSEUM: Andrews EXETER: See ROYAL ALBERT MEMORIAL MUSEUM, EXETER FREE PUBLIC MUSEUM, LIVERPOOL: See NATIONAL MUSEUMS AND GALLERIES ON MERSEYSIDE GALWAY, UNIVERSITY COLLEGE: Sec JAMES MITCHELL MUSEUM, UNIVERSITY COL- LEGE GALWAY GEOLOGICAL MUSEUM, INSTITUTE OF GEOLOGICAL SCIENCES, LONDON: See BRITISH GEOLOGICAL SURVEY, LONDON GEOLOGICAL SOCIETY, LONDON: See under BRITISH GEOLOGICAL SURVEY, KEYWO- RTH where the collection is housed. GEOLOGICAL SURVEY MUSEUM, LONDON: See BRITISH GEOLOGICAL SURVEY, LONDON GEOLOGICAL SURVEY OF IRELAND, DUBLIN: Morris 1988; Parkes and Sleeman; Wyse Jackson and Sleeman GEOLOGICAL SURVEY OF SCOTLAND, EDINBURGH: See BRITISH GEOLOGICAL SURVEY, EDINBURGH GLASGOW: See HUNTERIAN MUSEUM, UNIVERSITY OF GLASGOW GLASGOW ART GALLERY AND MUSEUM, KELVINGROVE [ formerly Kelvingrove Museum \ : Andrews; Benton; Campbell; Rolfe et at. GLOUCESTER CITY MUSEUM AND ART GALLERY : Torrens 1978a GRANT INSTITUTE, EDINBURGH : Morris 1988 GROSVENOR MUSEUM, CHESTER: Sarjeant HANCOCK MUSEUM, NEWCASTLE UPON TYNE: Boyd 1986; Boyd and Turner; Newman and Chatt-Ramsey HULL UNIVERSITY, GEOLOGY DEPART- MENT: Hodgkinson; Morris 1988 HUNTERIAN MUSEUM, UNIVERSITY OF GLASGOW: Benton; Cocks; Morris 1988; Rolfe et al. ; Sarjeant ISLE OF WIGHT: See MUSEUM OF ISLE OF WIGHT GEOLOGY JAMES MITCHELL MUSEUM, UNIVERSITY COLLEGE GALWAY: Pattison KEIGHLEY: See CLIFFE CASTLE MUSEUM, KEIGHLEY KESWICK MUSEUM AND ART GALLERY: Morris 1988 KILMARNOCK: See DICK INSTITUTE KINGSTON UPON HULL CITY MUSEUMS: Boyd 1983 LANCASHIRE MINING MUSEUM, SALFORD [formerly Salford Mining Museum ]: Sarjeant LAUNCESTON: See SOUTHGATE MUSEUM, LAUNCESTON LEICESTERSHIRE MUSEUM, ART GALLER- IES AND RECORDS SERVICE: Sarjeant LINNEAN SOCIETY COLLECTION, LONDON: Cocks LIVERPOOL: See NATIONAL MUSEUMS AND GALLERIES ON MERSEYSIDE LUDLOW MUSEUM: Morris 1988 MANCHESTER MUSEUM: Eagar and Preece; Hancock et al.-, Morris 1988; Newman and Chatt- Ramsey; Nudds 1992a, 1992b; Sarjeant MANX MUSEUM: Garrad MIDLAND GEOLOGICAL SOCIETY MUSEUM : Cocks MUSEUM OF IRISH INDUSTRY [ forerunner of National Museum of Ireland , Dublin ; and also held collections of Geological Survey of Ireland , Dublin ] : Morris 1988 MUSEUM OF ISLE OF WIGHT GEOLOGY: Radley MUSEUM OF NORTH DEVON, BARNSTAPLE [ formerly North Devon Athenaeum, Barnstaple]'. Butler; Morris 1988 NATIONAL MUSEUM OF IRELAND, DUBLIN : Cocks; Morris 1988; Wyse Jackson and Monaghan NATIONAL MUSEUM OF WALES, CARDIFF: Cocks; Morris 1988; Owens and Bassett; Sarjeant; Smith 1989, 1996 NATIONAL MUSEUMS AND GALLERIES ON MERSEYSIDE [incorporating Liverpool Museum ] : Phillips, P. W. NATIONAL MUSEUMS OF SCOTLAND. EDIN- BURGH [See also ROYAL SCOTTISH MUSEUM, EDINBURGH]: Smith 1989; Suth- erland NATURAL HISTORY MUSEUM, LONDON: Adams; Adams et al.-, Andrews; Benton; Cocks; Crane and Getty; Crawley; Duffin 1979; Ensom; 614 PALAEONTOLOGY, VOLUME 40 Hancock et al.\ Hodgkinson; Joysey; Lewis 1986, 1993; Morris 1980, 1988; Morris and Fortey; Newman and Chatt-Ramsey; Phillips, D. 1977, 1982a, 19826; Pyrah 1976, 1978; Smith 1989, 1996; Torrens 1978a, 19786, 1983; Torrens and Taylor; Tripp and Howells; Williams NEWCASTLE UPON TYNE: See HANCOCK MUSEUM, NEWCASTLE UPON TYNE NORTH DEVON ATHENAEUM, BARNS- TAPLE: See MUSEUM OF NORTH DEVON, BARNSTAPLE NORTHAMPTON: See CENTRAL MUSEUM AND ART GALLERY, NORTHAMPTON NORWICH : See CASTLE MUSEUM, NORWICH NOTTINGHAM UNIVERSITY GEOLOGY DEPARTMENT: Sarjeant OXFORD: See UNIVERSITY MUSEUM, OXFORD PETERBOROUGH CITY MUSEUM AND ART GALLERY: Cross 1975a, 19756 PORTSMOUTH : See UNIVERSITY OF PORTS- MOUTH ROCHDALE MUSEUM: Steward and Torrens ROWLEY’S HOUSE MUSEUM, SHREWSBURY: Sarjeant ROYAL ALBERT MEMORIAL MUSEUM, EXETER: Murray and Taplin ROYAL SCOTTISH MUSEUM, EDINBURGH [See also NATIONAL MUSEUMS OF SCOT- LAND, EDINBURGH]: Andrews; Baird; Benton; Morris 1988; Paton 1976, 1981 SALFORD MUSEUM OF MINING: See LANCA- SHIRE MINING MUSEUM, SALFORD SCARBOROUGH: See WOOD END MUSEUM OF NATURAL HISTORY, SCARBOROUGH SEDGWICK MUSEUM, CAMBRIDGE: Benton; Cocks; Crane and Getty; Morris 1988; Newman and Chatt-Ramsey; Smith 1989 SHEFFIELD CITY MUSEUM: Morris 1988; Riley SHERBORNE SCHOOL MUSEUM [All type spec- imens transferred to The Natural History Museum , London} -. Torrens 19786 SHREWSBURY: See ROWLEY’S HOUSE MUSEUM, SHREWSBURY SOUTHGATE MUSEUM, LAUNCESTON : Morris 1988 TORQUAY MUSEUM: Morris 1988 TRINITY COLLEGE, DUBLIN [GEOLOGICAL MUSEUM]: Cocks; Morris 1988; Nudds 1982a, 19826, 1982c, 1983, 1984, 1988, 1989; Tunnicliff; Wyse Jackson and Monaghan; Wyse Jackson and Sleeman TULLIE HOUSE MUSEUM AND ART GALL- ERY, CARLISLE: Cocks; Morris 1988 ULSTER MUSEUM, BELFAST: Tunnicliff UNIVERSITY OF CAMBRIDGE, ZOOLOGICAL MUSEUM: Adams; Joysey UNIVERSITY MUSEUM, OXFORD: Andrews; Cocks; Knell; Morris 1988; Newman and Chatt- Ramsey; Powell and Edmonds; Pyrah 1978 UNIVERSITY OF PORTSMOUTH [Formerly Portsmouth Polytechnic}'. Torrens and Taylor WELLCOME INSTITUTE [All type specimens trans- ferred to the Natural History Museum , London]'. Torrens 1983 WHITBY MUSEUM: Pyrah 1976 WOLLATON HALL NATURAL HISTORY MUSEUM, NOTTINGHAM: Morris 1988 WOOD END MUSEUM OF NATURAL HIS- TORY, SCARBOROUGH : Newman and Chatt- Ramsey WORCESTER CITY MUSEUM AND ART GALLERY: Cocks; Smith 1989 YORKSHIRE MUSEUM: Mancenido and Dambo- renea; Pyrah 1976, 1977, 1978, 1979a, 19796; Sarjeant UNPUBLISHED INFORMATION As a result of our soliciting information from many institutions while compiling this bibliography, a number of curators kindly commented on the status and availability of unpublished draft catalogues or on work in progress in identifying type, figured and cited specimens. These comments are quoted briefly here as they will be helpful to anyone suspecting that specimens they are attempting to trace might be in a particular museum. We have not seen the various manuscript catalogues quoted, and details from them are not included in the above indexes. Card index records in museums are not listed. BIRMINGHAM UNIVERSITY, LAPWORTH MUSEUM. ‘The type and figured catalogue was initiated by Laurie and ... is in manuscript only but is available upon request. The Type and Figured Collection currently comprises 2593 type, figured and cited specimens. A full computerised re-cataloguing programme is underway and a Type and Figured Catalogue will be published when this is completed’. [Paul Smith, Lapworth Museum, pers. comm. 1995] BOOTH MUSEUM OF NATURAL HISTORY, BRIGHTON. Typescript for type and figured catalogue available. [John Cooper, Booth Museum, pers. comm. 1995] DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 615 BRITISH GEOLOGICAL SURVEY, KEYWORTH, NOTTINGHAM. From 1984/5 all of BGS's type and figured palaeontological material from England and Wales has been housed at Keyworth. This includes all the material formerly housed at the Geological Museum, London, and the (mostly Carboniferous) collections held in Leeds from about 1959 to the time of the Keyworth move. Non-figured/cited material in the ‘Survey Collection' was relocated at the same time. Most of the Scottish type and figured material which was housed at Murchison House in Edinburgh is now also at Keyworth (for exceptions see Coprolite , 14, p. 2, 1994' [a newsletter of the Geological Curators' Group]). ‘ However, no comprehensive catalogue has been produced. The Type and Stratigraphic Collection amounts to approximately 250,000 specimens of which at least 30.000 are type/figured/cited specimens plus about 10,000 microfossils of similar status (Nudds 1994) . ...Although BGS has begun to enter this information on the database, it will be many years before time allows completion. In the meantime, type material is accessible through the collections themselves, through the registers and through annotated copies of most journals and books. There are various as-yet- unpublished lists of donors, purchases and so on which can be made accessible and which, in part, formed the basis of' [the] ‘entry for the Survey under “ Institute of Geological Sciences" Cleevely (1983). ... These are also useful in tracking down type material.’ [Steve Tunnicliff, BGS Keyworth, pers. comm. 1995] CENTRAL MUSEUM AND ART GALLERY, NORTHAMPTON. ‘Much work has been carried out on the type, figured and cited specimens...’ in the Beeby Thompson collection. It is hoped to produce a draft listing in the near future. [Angela Edgar, Central Museum, Northampton, pers. comm. 1995] HULL CITY MUSEUMS. Draft catalogue of the numerous type, figured and cited fossils lost in 1943 (prepared by P. J. Boylan) is available in typescript. [Heather Rayfield, Hull City Museums, pers. comm. 1995] NATIONAL MUSEUM OF IRELAND, DUBLIN. A number of typescript catalogues are ‘available for research workers, and copies of sections of them have also been made available for various scientists in advance of publication. All are intended for publication eventually...’ These include: Types, figured and cited specimens of fossils in the collection of Sir Richard John Griffith (1784—1878) including all specimens used by Frederick M’Coy in his works of 1844 and 1846. [for publication 1996] Types, figured and cited specimens of fossil plants in the collections of the National Museum of Ireland. Ditto . . . fossil cephalopods . . . Ditto ... Lower Palaeozoic invertebrates... Ditto . . . fossil fish ... Ditto ... fossil amphibians... Ditto... fossil reptiles... [Nigel Monaghan, pers. comm. 1995] SEDGWICK MUSEUM, UNIVERSITY OF CAMBRIDGE. Catalogue of complete collections stored on computer. ‘Complete hard-copy versions of the catalogue and a taxonomic index are generated periodically on microfiche' [R. B. Rickards, Geology Today , I. p. 184 ] SUPPLEMENTARY REFERENCES In addition to the publications listed above in the main Bibliography, valuable information on the location of numerous other palaeontological collections is included in various sources. The following list is not necessarily comprehensive, but does include the main compilations referring to material in the British Isles published over the past 20 years. Especially valuable in many of them is information on collectors and the present whereabouts of their collections, which may lead investigators towards potential repositories of hitherto untraced type specimens. Information in square brackets following the references gives a summary of the scope of the publication. Further useful supplementary sources include the continuing ‘Lost and Found’ series in The Geological Curator (indexed up to 1987 in Vol. 5, No. 2; and indexed for 1987 to 1994 in Vol. 6, No. 4) and the ‘Museum File' series published since 1985 in Geology Today (published six times a year by Blackwell Scientific Publications Ltd in association with the Geological Society of London and the Geologists’ Association); this latter series summarizes the scope of palaeontological and other collections in museums throughout the British Isles, generally with reference to major collectors and important publications. Because of the importance of adopting the highest standards of curatorial procedures in storing and maintaining type collections, we also include here a number of recent references that specifically cover procedures for curation of geological materials. 616 PALAEONTOLOGY, VOLUME 40 arnold-forster, K. 1989. The collections of the University of London: a report and survey of the museums , teaching and research collections administered by the University of London. London Museums Service, 41 pp. [Covers the history, development, scope (in very general terms), and current condition of collections in all London University colleges. The existence of some figured and cited specimens is noted]. bassett, M. G. 1975. Bibliography and index of catalogues of type, figured and cited fossils in museums in Britain. Palaeontology , 18, 753-773. (ed.). 1979. Curation of palaeontological collections. Special Papers in Palaeontology , 22, 1-280. [Covers the role of collections and curators, curatorial problems and procedures, techniques, and exhibits], bateman, j., McKenna, G. and timberlake, s. (eds). 1993. Natural science collections in south east Britain. Published for the South Eastern Collections Research Unit and AMMSEE by the Museum Documentation Association, 283 pp. [Lists collections under collectors’ name, and includes summaries of nature of collections, status, etc.]. brunton, c. h. c., besterman, t. p. and cooper, J. a. (eds). 1985. Guidelines for the curation of geological materials. Geological Society Miscellaneous Paper, No. 17, 209 pp. [Prepared by the Geological Curators’ Group, this loose-leaf manual covers acquisition, documentation, preservation, hazards and uses of geological material], child, r. e. (ed.). 1994. Conservation of geological collections. Proceedings of a conference held at the Welsh Folk Museum , National Museum of Wales , 4 November 1993. Conservation Monograph Series, Archetype Publications, London, for the National Museum of Wales and the Council of Museums of Wales, 65 pp. cleevely, r. j. 1983. World palaeontological collections. British Museum (Natural History), London, 365 pp. [Includes records of collections held in British and Irish institutions, together with information on the presence of type, figured, cited specimens and associated catalogues]. collins, c. (ed.). 1995. The care and conservation of palaeontological material. Butterworth-Heinemann Ltd, Oxford, 139 pp. crowther, p. r. 1990. Collection care and status material. 515-517. In briggs, d. e. g. and crowther, p. r. (eds). Palaeobiology : a synthesis. Blackwell, Oxford, xii-t-583 pp. — (ed.). 1992. Proceedings of a GCG meeting. Geology in Irish Museums, Trinity College Dublin, 21-22 June 1990. Geological Curator, 5, 261-300. [Summarizes the history and general content of collections in Trinity College Dublin, National Museum of Ireland, Geological Survey of Ireland, University College Galway]. davis, p. and brewer, c. (eds). 1986. A catalogue of natural science collections in north-east England. North of England Museums Service, 333 pp. [Gives list of collectors, plus summary of nature of collections and repositories], doughty, p. s. 1981. The state and status of geology in United Kingdom museums. Report on a survey conducted on behalf of the Geological Curators' Group. Geological Society of London, Miscellaneous Paper No. 13, 118 pp. [Comprehensive source of reference, based on questionnaires returned from c. 570 museums. Survey covered size, storage, named collections, condition, catalogue systems, geographical coverage. Appendices summarize scope of collections listed by collectors’ names, cross-referenced with index of museums covered], HANCOCK, E. G. and pettitt, c. w. (eds). 1981. Register of natural science collections in N.W. England. Manchester Museum, 188 pp. [Lists contents of geological, botanical, zoological collections in museums and related institutions, plus details of collectors, and existence of types]. hartley, m. m., norris, a., pettitt, c. w., riley, t. h. and stier, m. a. (eds). 1987. Register of natural science collections in Yorkshire and Humberside. Area Museum Service for Yorkshire and Humberside. [Gives list of collectors, summary of nature of collections and status, etc.]. international trust for zoological nomenclature 1985. International Code of Zoological Nomenclature, third edition, adopted by the XX General Assembly of the International Union of Biological Sciences. University of California Press, Berkeley and Los Angeles, 338 pp. nudds, j. r. 1994. Directory of British Geological Museums. Geological Society of London, Miscellaneous paper. No. 18, 141 pp. [Includes published catalogues, status of collections, etc.]. Paine, c. (ed.). 1993. Standards in the museum. 3. Care of geological collections. Museums and Galleries Commission, London, 57 pp. stace, h. e., pettitt, c. w. e. and waterston, c. d. 1987. Natural science collections in Scotland. National Museums of Scotland, 404 pp. -I- 8 microfiches. [Gives comprehensive summary of nature of collections, and status. All types of institution listed]. webby, b. D. (compiler) 1989. Fossil collections of the world: an international guide. International Palaeontological Association, Washington D.C. vi + 214 pp. [United Kingdom section covers national. DEISLER AND BASSETT: BIBLIOGRAPHY AND INDEX 617 county /city and university museums. Summarizes size and scope of collections and indicates published catalogues and type material], wiktor, J. and rydzewski, w. 1991 . Bibliography of catalogues of type specimens in [the] world's zoological and palaeozoological collections. Wroclaw University Press, Poland, 308 pp. Paperback [In English], [Publication not seen, but referred to in Cooper’s book review of Kabat and Boss in Geological Curator , 5, 372], Acknowledgements. We are most grateful to the many curators who have given us information about the collections for which they are responsible, especially data that supplement the published records. Dr R. M. Owens kindly discussed various queries throughout the course of our compilation. VALERIE K. DEISLER MICHAEL G. BASSETT Department of Geology Typescript received 8 July 1996 National Museum of Wales Revised typescript received 8 November 1996 Cardiff CF1 3NP, Wales, UK NOTE ADDED IN PROOF Subsequent to submission of the manuscript, two further publications have been issued which give additional information on collections and individual specimens relevant to this compilation. They are noted here for reference and completeness but their contents are not included in the above indexes. harper, d. a. t. and parkes, m. a. 1996. Geological Survey donations to the Geological Museum in Queen’s College Galway: 19th Century inter-institutional collaboration in Ireland. The Geological Curator, 6. 233-236. strachan, i. 1996. A bibliographic index of British graptolites (Graptoloidea). Part 1. Monograph of the Palaeontographical Society , 150 (600), 1-40. NOTES FOR AUTHORS The journal Palaeontology is devoted to the publication of papers on all aspects of palaeontology. Review articles are particularly welcome, and short papers can often be published rapidly. A high standard of illustration is a feature of the journal. Four parts are published each year and are sent free to all members of the Association. Typescripts should conform in style to those already published in this journal, and should be sent (with a disk, if possible) to the Secretary of the Publications Committee, Dr R. M. Owens, Department of Geology, National Museum of Wales, Cardiff CF1 3NP, UK, who will supply detailed instructions for authors on request (these are published in Palaeontology 1996, 39, 1065-1075). Special Papers in Palaeontology is a series of substantial separate works conforming to the style of Palaeontology. SPECIAL PAPERS IN PALAEONTOLOGY In addition to publishing Palaeontology the Association also publishes Special Papers in Palaeontology. Members may subscribe to this by writing to the Membership Treasurer: the subscription rate for 1997 is £55 00 (U.S. $120) for Institutional Members, and £20 00 (U.S. $36) for Ordinary and Student Members. A single copy of each Special Paper is available on a non-subscription basis to Ordinary and Student Members only , for their personal use, at a discount of 25 per cent, below the listed prices : contact the Marketing Manager. Non-members may obtain Nos 35-56 (excluding 44) at cover price from Blackwell Publishers Journals, P.O. Box 805, 108 Cowley Road, Oxford OX4 1FH, UK, and older issues from the Marketing Manager. For all orders of Special Papers through the Marketing Manager, please add £1 -50 (U.S. $3) per item for postage and packing. PALAEONTOLOGICAL ASSOCIATION PUBLICATIONS Special Papers in Palaeontology For full catalogue and price list, send a self-addressed, stamped A4 envelope to the Marketing Manager. Numbers 2-47, excluding 44, are still in print and are available together with those listed below: 48. (for 1992): Contributions to acritarch and chitinozoan research. Edited by s. G. molyneux and k. j. dorning. 139 pp., 28 text-figs , 22 plates. Price £40 (U.S. $80). 49. (for 1993): Studies in palaeobotany and palynology in honour of Professor W. G. Chaloner, F.R.S. Edited by m. e. collinson and a. c. scott. 187 pp., 38 text-figs, 27 plates. Price £50 (U.S. $100). 50. (for 1993): Turanian ammonite faunas from central Tunisia, by g. r. chancellor, w. j. Kennedy and j. m. Hancock. 1 18 pp., 19 text-figs, 37 plates. Price £40 (U.S. $80). 51. (for 1994): Belemnitella from the Upper Campanian and Lower Maastrichtian Chalk of Norfolk, England, by w. k. Christensen. 84 pp., 22 text-figs, 9 plates. Price £35 (U.S. $70). 52. (for 1994): Studies on Carboniferous and Permian vertebrates. Edited by a. r. milner. 148 pp., 51 text-figs, 9 plates. Price £45 (U.S. $90). 53. (for 1 995) : Mid-Dinantian ammonoids from the Craven Basin, north-west England, by N. j. riley. 87 pp., 51 text-figs, 8 plates. Price £40 (U.S. $80). 54. (for 1995): Taxonomy and evolution of Llandovery biserial graptoloids from the southern Urals, western Kazakhstan, by t. n. koren’ and r. b. rickards. 103 pp., 23 text-figs, 1 4 plates. Price £40 (U.S. $80). 55. (for 1996): Studies on early land plant spores from Britain. Edited by c. i. cleal. 145 pp., 23 text-figs, 28 plates. Price £45 (U.S. $90). 56. (for 1996): Fossil and Recent eggshell in amniotic vertebrates: fine structure, comparative morphology and classification, by k. e. Mikhailov. 80 pp., 21 text-figs, 15 plates. Price £35 (U.S. $70). Field Guides to Fossils and Other Publications These are available only from the Marketing Manager. Please add £1-00 (U.S. $2) per book for postage and packing plus £1-50 (U.S. $3) for airmail. Payments should be in Sterling or in U.S. dollars, with all exchange charges prepaid. Cheques should be made payable to the Palaeontological Association. 1. (1983): Fossil Plants of the London Clay, by m. e. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $16) (Members £6 or U.S. $12). 2. (1987): Fossils of the Chalk, compiled by E. owen; edited by a. b. smith. 306 pp., 59 plates. Price £1L50 (U.S. $23) ( Members £9-90 or U.S. $20). 3. (1988): Zechstein Reef fossils and their palaeoecology, by n. hollingworth and t. pettigrew. iv+75 pp. Price £4-95 (U.S. $10) (Members £3-75 or U.S. $7-50). 4. (1991): Fossils of the Oxford Clay, edited by d. m. martill and i. d. Hudson. 286 pp., 44 plates. Price £15 (U.S. $30) (Members £ 1 2 or U.S. $24). 5. (1993): Fossils of the Santana and Crato Formations, Brazil, by d. m. martill. 159 pp., 24 plates. Price £10 (U.S. $20) (Members £7.50 or U.S. $15). 6. (1994): Plant fossils of the British Coal Measures, by c. i. cleal and b. a. thomas. 222 pp., 29 plates. Price £12 (U.S. $24) (Members £9 or U.S. $18). 7. (1996): Fossils of the upper Ordovician, edited by d. a. t. harper and a. w. owen. 312 pp., 52 plates. Price £16 (U.S. $32) (Members £12 or U.S. $24). 1985. Atlas of Invertebrate Macrofossils. Edited by i. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii +241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24-95. © The Palaeontological Association, 1997 Palaeontology VOLUME 40 • PART 2 CONTENTS Postcranial morphology and locomotor behaviour of two early Eocene miacoid carnivorans, Vulpavus and Didymictis RONALD E. HEINRICH and KENNETH D. ROSE 279 Anatomy and relationships of the pareiasaur Pareiasnchus nasicornis from the Upper Permian of Zambia M. S. Y. LEE, C. E. GOW and J. W. KITCHING 307 Early Cambrian brachiopods from North Greenland LEONID POPOV, LARS E. HOLMER, ALBERT J. ROWELL and john s. peel 337 Late Jurassic brachiopods from north-east Iran M. H. adabi and D. V. AGER 355 A new mitrate from the lower Ordovician of southern France MARCELLO RUTA 363 Mode of life of the Middle Cambrian eldonioid lophophorate Rotadiscus JERZY DZIK, ZHAO YUANLONG and Z H U MAOYAN 385 Late Ordovician trilobites from southern Thailand RICHARD A. FORTEY 397 Bubble-headed trilobites, and a new olenid example R. A. FORTEY and R. M. OWENS 451 A revision of the large lagomerycid artiodactyls of Europe BEATRIZAZANZA and LEONARD GINSBURG 461 A new transitional myalinid bivalve from the Lower Permian of west Texas CHRISTOPHER A. McROBERTS and NORMAN D. NEWELL 487 Early Jurassic brachiopods from Gibraltar, and their Tethyan affinities ELLIS F. OWEN and EDWARD P. F. ROSE 497 Life histories of some Mesozoic encrusting cyclostome bryozoans frankk. McKinney and Paul d. taylor 515 The diet of the early Toarcian ammonite Harpoceras falcifemm MANFRED JAGER and RENE FRAAYE 557 Late Devonian winged preovules and their implications for the adaptive radiation of early seed plants N. P. ROWE 575 Bibliography and index of catalogues of type, figured and cited fossils in museums in Great Britain and Ireland (Supplement 1975-1996) VALERIE K. DEISLER and MICHAEL G. BASSETT 597 Printed in Great Britain at the University Press, Cambridge ISSN 0031-0239 HECKMAN BINDERY INC. AUG 98 Bound -To-PIc^ ^MANCHESTER libraries 3 9088 01375 71 on